Brazil: Capital Flight, Illicit Flows, and

Macroeconomic Crises, 1960-2012
Dev Kar
September 2014
Brazil: Capital Flight, Illicit Flows, and
Macroeconomic Crises, 1960-2012

Dev Kar
1
September 2014
Global Financial Integrity Wishes to Thank
the Ford Foundation for Supporting this Project
1.
Dev Kar is Global Financial Integrity’s Chief Economist, having formerly served as a Senior Economist at the International Monetary
Fund. Brian LeBlanc assisted with the data analysis, and Joshua Simmons contributed to the policy analysis. Raymond Baker, Christine
Clough, Clark Gascoigne, Taylor Le, Channing May, and Melissa O’Brien also supported this project.
iii Brazil: Capital Flight, Illicit Flows, and Macroeconomic Crises, 1960-2012
We are pleased to present here our report, Brazil: Capital Flight, Illicit Flows, and Macroeconomic
Crises, 1960-2012.
Illicit fnancial outfows averaged US$14.7 billion per year for the period from 2000 to 2009. For the
period from 2010 to 2012, illicit fnancial outfows increased to an average of US$33.7 billion per
year. These outfows constitute about 1.5 percent of Brazil’s growing GDP for both of these periods.
In terms of total magnitude, the country is seventh among developing countries, all of which suffer
from this phenomenon.
GFI’s analysis is based on data fled by Brazil with the International Monetary Fund and the World
Bank, enabling a breakdown of unrecorded outfows into balance of payments leakages and trade
misinvoicing. Balance of payments leakages have across the decades generally been on the order
of 10 to 20 percent of the total, meaning that trade misinvoicing generally accounts for 80 to 90
percent of the drainage of capital from the country.
For many years we have observed a hesitancy in Brazil to address problems of capital fight and
illicit outfows, as many scholars and offcials believed that the nation’s strong sense of patriotism
and burgeoning economy worked to dampen this possibility. It is, however, real and merits serious
attention by policymakers.
GFI has produced earlier in-depth studies of illicit fnancial outfows affecting India, Russia,
Mexico, and the Philippines and more limited analyses of similar issues affecting China, Ghana,
Kenya, Tanzania, Uganda, and Mozambique. With this study of Brazil, we utilize for the frst time a
comprehensive structural equations model (SEM) to examine factors that infuence licit and illicit
fows from the country. We fnd that the illicit outfows—comprising some 68 percent of the total—
make up the largest component of the outfows of capital. Furthermore, we fnd that the dominant
factor correlating to both of these components of outfows is the impact of the underground
economy, which both drives and is driven by illicit outfows, confrming the importance of dealing
with this aspect of the economy and its roots in governance concerns.
These observations lead to the conclusion that more citizens operating in the underground
economy need to be brought into the framework of the legal economy, and that central to
accomplishing this goal is the curtailment of trade misinvoicing. Brazil has a distinctive approach
to the issue of misinvoicing, particularly that part of the problem encountered with multinational
iv Global Financial Integrity
corporations—abusive transfer pricing. In particular, subjecting transactions with tax havens to
special scrutiny has restrained exaggerated charges for intangibles and services. The government
should do much more to curtail both the under-pricing of exports and the over-pricing of imports
through additional proactive deterrence measures, rather than retroactive punishment.
GFI thanks Dev Kar, chief economist, Brian LeBlanc, associate economist, and Joshua Simmons,
policy counsel, for their outstanding work on this analysis. GFI also thanks the Ford Foundation
and its former program offcer, Leonardo Burlamaqui, Brazilian himself, for years of generous
contributions both to this work and to earlier studies.
Raymond W. Baker
President
September 2014
v Brazil: Capital Flight, Illicit Flows, and Macroeconomic Crises, 1960-2012
Contents
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
Executive Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
I. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
II. Capital Flight and Illicit Flows in the Context of Macroeconomic Crises . . . . . . . . . . . . . . . . . . . . 3
i. Estimating Capital Flight and Illicit Financial Flows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
ii. Capital Flight and Illicit Flows in the Historical Context . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
a. Overall Volume and Pattern . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
b. Overview of Economic History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1. Broad Capital Flight and Macroeconomic Crises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2. The Nature and Scale of Illicit Financial Flows from Brazil . . . . . . . . . . . . . . . . . . . . . . . . 11
III. A Model of Capital Flight and Illicit Financial Flows from Brazil . . . . . . . . . . . . . . . . . . . . . . . . . . 15
i. Government Sector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
ii. Money Supply Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
iii. Formation of Prices. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
iv. Real Sector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
v. Underground Economy, Illicit Flows, and Capital Flight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
vi. The Complete Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
vii. Results of Dynamic Simulation of the SEM. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
viii. Inequality, Capital Flight, and Illicit Flows. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
IV. The Legal and Policy Environment in Brazil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
i. Customs, Trade, and Tax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
ii. Transparency and Governance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
iii. Financial Regulation and Governance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
iv. Policy Recommendations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
a. Customs and Trade Reform. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
b. Financial Transparency and Governance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
c. Effective Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
V. Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Appendix I. Capital Flight and Illicit Financial Flows by Year . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
Appendix II. The Components of Brazil’s Trade Misinvoicing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
Appendix III. Illicit Flows to GDP & Trade . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
Appendix IV. Estimating Brazil’s Underground Economy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
About the Author . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

vi Global Financial Integrity
Charts and Tables
Chart 1. Brazil: Broad Capital Flight and Illicit Outfows, 1960-2009 . . . . . . . . . . . . . . . . . . . . . . . . . 5
Chart 2. Brazil: Broad Capital Flight and Illicit Outfows, 1960-2012 . . . . . . . . . . . . . . . . . . . . . . . . . 6
Table 1. Brazil: Decennial Developments in Capital Flight and Illicit Financial Flows from Brazil. . . . 7
Chart 3. Brazil: Capital Flight and Macroeconomic Crises, 1965-2012 . . . . . . . . . . . . . . . . . . . . . . 10
Chart 4. Brazil: Illicit Financial Outfows and the Underground Economy, 1960-2009. . . . . . . . . . . 13
Table 2. Brazil: Components of the Structural Equations Model and
How They Impact Target Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
Table 3. Multiple Regression Analysis: Links between Economic Growth,
Income Inequality, and Capital Flight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Appendix I: Table 1. Brazil: Broad Capital Flight and Illicit Financial Flows, 1960-2012. . . . . . . . . . 35
Appendix II: Table 2. Brazil: The Components of Trade Misinvoicing, 1960-2012 . . . . . . . . . . . . . . 36
Appendix III: Table 3. Brazil: Illicit Flows to GDP and Trade, 1960-2012. . . . . . . . . . . . . . . . . . . . . . 37
Appendix IV: Chart 4. Underground Economy in Brazil, Decadal Averages, 1960-2012. . . . . . . . . 39
Appendix IV: Table 4. Underground Economy in Brazil, Decadal Averages, 1960-2012 . . . . . . . . . 39
vii Brazil: Capital Flight, Illicit Flows, and Macroeconomic Crises, 1960-2012
Abstract
This study examines capital fight and illicit fnancial fows from Brazil against a backdrop of
macroeconomic crises. Over the 53 year period covered in this study, the Brazilian economy
experienced high infation, hyperinfation, large fscal defcits, and crushing external debt, which
led to debt reschedulings and prolonged recessions. The paper sheds light on the behavior of
capital fight in response to such shocks and how illicit fows from the country move in tandem
with its underground economy. Given the existing gap in academic literature, we use a full-scale
structural equations model to analyze the link between broad capital fight and illicit fows on the
one hand and macroeconomic and governance-related drivers on the other. Specifcally, we model
fscal operations, monetary policy, price developments, GDP, and capital formation along with
the behavior of the underground economy and capital fight in order to study their interactions. A
salient model result is that, while illicit fows both drive and are driven by the underground economy,
broad capital fight is driven by macroeconomic drivers such as monetary policy and investment
conditions as well as illicit fows. Informed by the results of the model, we conclude with policy
recommendations to curtail the cross-border transfer of such capital.
viii Global Financial Integrity
ix Brazil: Capital Flight, Illicit Flows, and Macroeconomic Crises, 1960-2012
Executive Summary
This is a study of illicit fnancial fows and capital fight (consisting of both licit as well as illicit
capital) from Brazil, which we have undertaken for three main reasons. First, existing studies on
capital fight from Brazil are dated. Second, there are no studies that focus on outfows of illicit
capital from Brazil, let alone over a long time span. Finally, Brazil is a large exporter of illicit capital.
Global Financial Integrity’s latest annual report, Illicit Financial Flows from Developing Countries:
2002-2011, published in December 2013, found that the country illegally transferred abroad on
average US$19.3 billion per annum over the decade ending 2011, the seventh highest of such
outfows from the developing world.
Outfows of illicit capital totaled US$401.6 billion from 1960 through 2012. Illicit outfows increased
from an annual average of 1.49 percent of GDP in the 1960s to 1.71 percent of GDP in the 1980s
before receding to 1.54 percent of GDP in the last decade ending 2009. Outfows have continued
at roughly 1.47 percent of GDP in the most recent three-year period, 2010-2012, for which data are
available. Most illicit outfows occur through trade misinvoicing rather than via balance of payments
leakages. The deliberate under-invoicing of exports rather than the over-invoicing of imports is the
preferred method to transfer illicit funds from Brazil.
In terms of magnitude, Brazil lost a total of US$590.2 billion during the period (or about 2.2 percent
of GDP on average) through broad capital fight, which consists of illicit as well as licit funds. Capital
fight increased sharply from the 1960s through the 1990s, although the pace of outfows declined
somewhat over the last decade ending 2009. There is no doubt that serious macroeconomic shocks
related to hyperinfation and near-debt defaults triggered the continued increase in capital fight in
the 1990s. Starting at about 2.6 percent of GDP on average during the 1960s, capital fight fell to 1.9
percent of GDP in the last decade ending 2009 before increasing to 2.1 percent of GDP in the last
three years, 2010-2012.
We developed models to estimate the size of the underground economy and to explore the factors
driving both capital fight and illicit fows from Brazil over the period 1965-2011. One of the most
interesting fndings is that illicit fnancial fows both drive and are driven by Brazil’s underground
economy. Moreover, the models confrm that the underground economy had a signifcant negative
impact on investment, implying that, as the underground economy grew, it tended to divert resources
away from the offcial economy. Likewise, we found that even broad capital fight, consisting of a mix
of licit and illicit fows, is driven mainly by governance-related factors such as illicit fows. This is not
surprising given that, on average, illicit fows comprise some 68 percent of capital fight.
We found that broad capital fight behaved in a more predictable manner than illicit outfows in
response to macroeconomic crises. This is to be expected because the licit component of broad
capital fight tends to be more sensitive to macroeconomic shocks than the illicit component, where
x Global Financial Integrity
the primary motivation is sheltering illicit assets from regulatory scrutiny and confscation even in
the best of times. In general, capital fight seems to increase in the immediate aftermath of a crisis,
perhaps in proportion to the severity of the crisis.
Econometric model tests provided several insights into how macroeconomic conditions and the
overall state of governance impact both capital fight and illicit fows, as well as how they impact the
“above-ground,” or offcial, economy. The salient fndings can be summarized as follows.
Capital fight and illicit fnancial fows tend to drive each other. This is the frst study where a defnite
link between the two has been established through the use of a structural equations model. We
found that a 1.0 percent increase in illicit fnancial fows is correlated to a 0.83 percent increase in
broad capital fight.
We also examined the link between economic growth, income inequality, and capital fight.
While the lack of an unbroken series on the Gini coeffcient prevented its inclusion in our model,
regression analysis with a shorter time period (1970-2011) showed that worsening income inequality
also seems to drive capital fight, although the relationship is only signifcant at the 90 percent level.
A possible explanation is that rising income inequality implies a larger number of high net worth
individuals (HNWIs). It is the HNWIs rather than the common man that can fnance capital fight and
take advantage of the world’s shadow fnancial system to shelter wealth.
Drawing upon the results of the model, we examine the legal and policy environment in Brazil, and
conclude with suggested policy measures to curtail capital fight and illicit fows from Brazil. Our
fnding that illicit fows through trade misinvoicing comprise the largest proportion of capital fight
from Brazil suggests that curbing capital fight will require strong customs and tax enforcement
and oversight. Brazil has also long struggled with corruption, and our fnding of the persistent size
of Brazil’s underground economy—38.9 percent of the offcial economy per year on average over
the period of the study—suggests that the country faces much broader governance issues. Finally,
while Brazil has made great strides in recent years towards bringing its anti-money laundering
regime in line with international standards, these legal changes have not necessarily been
accompanied by effective enforcement.
No set of policy changes is capable of completely eliminating illicit fnancial fows or capital fight,
but we recommend several measures designed to substantially curtail such fows, guided by
two main principles: greater transparency in domestic and international fnancial transactions,
and greater cooperation between governments to shut down the channels through which illicit
money fows. These include taking stronger legal measures against trade misinvoicing, instituting
transparency of company ownership, and building the technical and human capacity needed
to effectively utilize the data that will be shared under emerging tax information exchange
arrangements.
xi Brazil: Capital Flight, Illicit Flows, and Macroeconomic Crises, 1960-2012
Overall, Brazil has an established fnancial infrastructure, a strong commitment to democratic
governance, and many of the laws and procedures needed to curb illicit fnancial fows and rein in
the underground economy already in place. However, these advantages must be coupled with the
capacity and political will to fully implement and enforce such measures. Curtailing illicit fnancial
fows must become a priority throughout the Brazilian government.


xii Global Financial Integrity
1 Brazil: Capital Flight, Illicit Flows, and Macroeconomic Crises, 1960-2012
I. Introduction
There are a number of reasons why a study of capital fight and illicit fows from Brazil is important.
Global Financial Integrity’s December 2013 report, Illicit Financial Flows from Developing Countries:
2002-2011, found that Brazil was the world’s seventh largest exporter of illicit capital, with
outfows averaging US$19.3 billion per annum over the decade ending 2011.
2
The country also
has a checkered economic history, ranging from fast economic growth to stagnation and even
contraction. Recessionary episodes were typically accompanied by severe macroeconomic crises
such as hyperinfation, external debt default, currency and exchange crisis, or stagfation wherein
tepid growth co-existed with high infation. There are hardly any studies that examine how capital
fight and illicit fows from Brazil behaved in response to various macroeconomic crises over time.
Finally, the study is notable given the paucity of academic literature on the interaction between
capital fight and illicit fows, as well as on how such outfows impact the offcial economy.
We develop a full-scale structural equations model (SEM) in order to study the behavior of broad
capital fight and illicit fows in the context of Brazil’s macroeconomic history. The SEM seeks to
capture the interactions of the offcial economy and broad capital fight as well as illicit fows. There
are two reasons why we need to use both measures of capital outfows in the case of Brazil. For
one, there has been a massive structural transformation of the Brazilian economy over more than
fve decades as extensive controls were dismantled in fts and starts towards greater economic
liberalization. As a result, outfows that were once considered illegal due to exchange controls
became legitimate due to capital account liberalization over time. For another, as an International
Monetary Fund (IMF) study noted, capital fight itself is “a somewhat elusive concept” requiring
us to distinguish between illegal outfows and those that are “normal,” in that they take place due
to considerations related to portfolio diversifcation and return maximization.
3
A singular focus on
capital fows that are strictly illicit would not only ignore structural changes in the economy, but it
would miss signifcant outfows due to normal investors’ concerns.
One of the hypotheses we will test is whether outfows of legitimate capital tend to be more strongly
linked to macroeconomic drivers compared to outfows of purely illicit capital. We say “tend”
because the macroeconomic conditions that drive capital fight typically vary from one country to
another. For example, it is hard to fnd a clear link between fscal defcits and capital fight because
the threshold defcits that could trigger outfows of capital (due to a fear of future tax increases
arising from increased defcits) may vary depending on the sources of defcit fnancing and what
economic agents consider to be excessive. Moreover, capital outfows due to covered interest
differentials may be larger in countries with more integrated capital markets than in countries whose
2.
Dev Kar and Brian LeBlanc. Illicit Financial Flows from Developing Countries: 2002-2011 (Washington, DC: Global Financial Integrity,
2013), 24.
3.
Michael Deppler and Martin Williamson, “Capital Flight: Concepts, Measurement, and Issues,” in Staff Studies for the World Economic
Outlook SM/87/24 (Washington, DC: International Monetary Fund, 23 January 1987), 39.
2 Global Financial Integrity
capital markets are less integrated due to tighter controls on capital fows.
4
We intend to shed light
on the particular drivers of capital fight and illicit fows in the Brazilian context.
The paper is organized as follows. Section II presents a brief methodological overview of how
broad capital fight and illicit fnancial fows are estimated, followed by a short discussion of Brazil’s
economic history. In this context, we will explore the behavior of capital fight and illicit fows in the
context of Brazil’s macroeconomic crises. We will then develop an SEM in Section III laying out
the theoretical basis for each structural equation and discuss the main fndings arising from model
simulations. In Section IV, we present a discussion of the policy measures needed to curtail such
capital outfows. The main conclusions of this study are summarized in Section V.
4.
A covered interest differential is defned as the difference in interest rates between two countries after taking account of the cost of
using a forward contract to cover or eliminate the investor’s exposure to exchange rate risks over the time period during which the
foreign investment matures.
3 Brazil: Capital Flight, Illicit Flows, and Macroeconomic Crises, 1960-2012
II. Capital Flight and Illicit Flows in the
Context of Macroeconomic Crises
i. Estimating Capital Flight and Illicit Financial Flows
Economists have long recognized that capital fight is a diffcult concept to measure. The early
works of Cuddington,
5
Cumby and Levich,
6
Dooley,
7
and other researchers point out such
diffculties and suggest alternative ways of capturing what are predominantly unrecorded capital
fows in both directions. The focus here is not to provide an overview of the academic literature on
various measures (as other researchers have already done so) but to draw a clear methodological
distinction between capital fight and illicit fows.
Estimates of broad capital fight presented in this study are based on the World Bank Residual (WBR)
method adjusted for trade misinvoicing. The WBR method was developed at the World Bank in 1985
and has been used extensively as a measure of broad capital fight that includes both licit and illicit
capital. Essentially, the WBR measure estimates the gap between a country’s recorded source of
funds and recorded use of funds. Source of funds consists of new external loans (estimated by adding
together the change in the stock of public and publicly guaranteed debt as well as private non-
guaranteed debt) plus net foreign direct investment (FDI). Typically, for a developing country like Brazil,
which receives more FDI than it invests abroad, net FDI infows would supplement the country’s source
of funds. Use of funds comprises fnancing of current account defcits as well as the additions to reserve
holdings. Should the country have a current account surplus or should it draw down reserves rather
than add to such holdings, such transactions would entail a negative use (i.e., add to source of funds).
The relationship between capital fight and illicit fows can be derived from the balance of payments
identity as enumerated by Claessens and Naudé.
8
Using their nomenclature, let A be the current account
balance, B represent net equity fows (including net foreign direct investment and portfolio investment),
C the other short-term capital of other sectors, D the portfolio investments involving other bonds, E the
change in deposit money banks’ foreign assets, F the change in reserves of the central bank, G the net
errors and omissions (NEO), and H the change in external debt. Then, the balance of payments identity is:
A + B + C+ D + E + F + G + H = 0
Or, C + D + E + G = -(A + B + F + H)

which implies that recorded, and therefore legal, private capital fows (C + D + E) plus net errors
and omissions (G) must equal the negative of the sum of the current account balance (A), net equity
5.
John T. Cuddington, “Capital Flight: Estimates, Issues, and Explanations,” in Princeton Studies in International Finance (Princeton, NJ:
International Finance Section, Dept. of Economics, Princeton University, 1986).
6.
Robert Cumby and Richard Levich, “On the Defnition and Magnitude of Recent Capital Flight,” in Capital Flight and Third World Debt,
eds. D. Lessard and J. Williamson (Washington, DC: Institute for International Economics, 1987).
7.
Michael P. Dooley, Capital Flight: A Response to Differences in Financial Risks, IMF Staff Papers 35 (Washington, DC: International
Monetary Fund, 1988).
8.
Stijn Claessens and David Naudé, Recent Estimates of Capital Flight, Policy Research Working Paper Series 1186 (Washington, DC:
Debt and International Finance Division, International Economics Department, The World Bank, 1993), 3-5.
4 Global Financial Integrity
fows (B), change in reserves (F), and the change in external debt (H). The right hand side of the above
equation is the WBR equation.
Economic sources and methods cannot capture illicit fnancial fows in a comprehensive manner.
The diffculty arises from the fact that we are in part trying to capture fnancial fows generated from
purely illegal activities such as drug traffcking, bribery and kickbacks related to corruption, and arms
or human smuggling, which are typically designed to evade detection by regulatory authorities and
law enforcement. Such activities are often settled in cash, so that the parties to the illegal transaction
cannot be traced. Hence, gap analysis of offcially recorded data has inherent limitations in capturing
illegal transactions. Nevertheless, in accordance with the methodology adopted by other researchers,
we replace the WBR estimates by the net errors and omissions (NEOs) of the balance of payments.
As past researchers such as Cumby and Levich have noted, the NEO is the main part of the Hot Money
Narrow (HMN) method, which also includes recorded short-term capital fows of the private sector.
9
An
IMF study points out that “…errors and omissions [in the balance of payments accounts] are implicitly
attributed in their entirety to capital transactions whose net value can be attributed to capital fight.”
10

Because NEOs refect unrecorded transactions, we attribute them to illicit fows. The HMN estimates
are also adjusted for trade misinvoicing to derive total illicit fnancial fows. Hence, the trade misinvoicing
component is common to both broad capital flight as well as illicit financial flows.
The NEO (G) can be derived from the balance of payments identity quite simply as follows:
G = -(A + B + F + H) – C – D – E
Or, G = -(A + B + F + H) – (C + D + E)

In other words, the NEO represents the difference between broad capital fight (as measured by the WBR
method) and licit private capital fows that are recorded by balance of payments compilers. What we are
doing is taking out licit capital fows from a mix of licit and illicit capital fows captured by the WBR method,
leaving us with illicit capital fows (or net errors and omissions). Both WBR and NEO (G) are supplemented by
estimates of trade misinvoicing to yield broad capital fight and illicit fnancial fows used in this study.
We shall see in the following section that while illicit fows are in general smaller than broad capital
fight (as intuitively they should be), the net entries in each component of the balance of payments
identity imply that they need not always be.
ii. Capital Flight and Illicit Flows in the Historical Context
a. Overall Volume and Pattern
Over the 53-year period 1960-2012, Brazil lost a total of US$590.2 billion through broad capital fight, of
which about US$401.6 billion was through illicit outfows (Table 1). These outfows represent around 2.2
9.
Cumby and Levich, “Defnition and Magnitude,” 12.
10.
Deppler and Williamson, “Concepts, Measurement, and Issues,” 42.
5 Brazil: Capital Flight, Illicit Flows, and Macroeconomic Crises, 1960-2012
percent and 1.5 percent of GDP on average, respectively. The implication is that, on average, illicit
outfows comprise around 68 percent of total broad capital fight, which includes licit capital outfows
as part of “normal” portfolio diversifcation. Again, this observation is subject to the caveat that licit
and illicit fnancial outfows need not necessarily add to broad capital fight.
The reason they may not is related directly to the balance of payments accounting framework, where
each entry is a net of infows minus outfows. For instance, an entry related to infows of foreign
direct investment may well be a net of loans minus repayments from the local subsidiary to its parent
company abroad. Hence, even a credit item is a net of fnancial fows in both directions. In such a
system, when we do not have the data on gross fows in both directions, it is possible that the sum
of the net items may not be equal to the gross result obtained by summing up the components. In
general, however, it is reasonable to say that illicit outfows comprise the major share (68 percent) of
total capital fight from Brazil. Given the signifcant difference between broad capital fight and illicit
fows, we model both of them in order to determine whether one drives the other.
The volume of broad capital fight increased from the 1960s through the 1990s at a blistering pace
as a result of macroeconomic instability such as high and highly variable infation and hyperinfation
(Chart 1). Outfows continued to increase during the 2000s but at a slower pace, although, during
the last three years, 2010-2012, they again picked up pace.
Chart 1. Brazil: Broad Capital Flight and Illicit Outflows, 1960-2009
(millions of nominal U.S. dollars)

3,097
18,230
46,713
85,720
146,686
3,979
29,899
65,940
159,056
184,135
0
20,000
40,000
60,000
80,000
100,000
120,000
140,000
160,000
180,000
200,000
1960-1969 1970-1979 1980-1989 1990-1999 2000-2009
M
i
l
l
i
o
n
s

o
f

U
.
S
.

D
o
l
l
a
r
s

Illicit Financial OuAlows Broad Capital Flight (Comprising Both Illicit and Licit Financial OuAlows)
6 Global Financial Integrity
The latest surge in broad capital fight (not shown in the chart) is likely to be a result of large
outfows of licit capital due to the liberalization of the Brazilian capital account as well as due to
ongoing fnancial globalization, which tends to integrate capital markets.
The pattern of illicit outfows has also increased at a rapid pace in line with broad capital fight.
There was a six-fold increase in such outfows from the 1960s to the 1970s, although the rate of
increase declined steadily over the decades; from the 1970s to the 1980s, cumulative outfows
grew by about two and a half times; from the 1980s to the 1990s, the increase was just 1.8 times;
the increase in the last decade from the 1990s to the 2000s decelerated to 1.7 times. These broad
patterns in the fnancial outfows under capital fight and IFFs are captured in Chart 1.
Chart 2. Brazil: Broad Capital Flight and Illicit Outflows, 1960-2009
(percent of GDP)
1.49% 1.50%
1.71%
1.38%
1.54%
2.64%
2.46%
2.41%
2.56%
1.93%
0.00%
0.50%
1.00%
1.50%
2.00%
2.50%
3.00%
1960-1969 1970-1979 1980-1989 1990-1999 2000-2009
P
e
r
c
e
n
t

o
f

G
D
P

Illicit Financial Ou@lows Broad Capital Flight (Comprising Both Illicit and Licit Financial Ou@lows)

In terms of GDP, broad capital fight declined somewhat from the 1960s through the 1980s,
increased marginally in the 1990s, but fell signifcantly in the last decade ending 2009 (Chart 2). In
other words, while capital fight in relation to GDP fell signifcantly from the 1990s to the 2000s, illicit
outfows in relation to GDP increased in the last decade over the 1990s, although the increase was
not suffcient to reach the peak from the 1980s (see Chart 2).
b. Overview of Economic History
During the 1960s and 1970s, Brazil had one of the world’s fastest growing economies. Growth
was led by expansion of exports. But the root cause of Brazil’s economic diffculties in the 1960s
7 Brazil: Capital Flight, Illicit Flows, and Macroeconomic Crises, 1960-2012
was infation, measured by the general price index, which is a weighted average of the cost of
living, wholesale prices, and the cost of construction in Rio de Janeiro. In fact, chronic infation
accompanied impressive economic growth during this decade. The rate of infation peaked at
slightly over 90 percent in 1964 before decelerating towards the end of the decade. Monetary
and credit expansion was signifcantly driven in large part by large cash defcits of the central
government. The fscal imbalance did not arise due to shortfalls in revenue collections but by
the growing operational defcits of state-owned enterprises and autonomous agencies and the
extension of various subsidies. Wage increases in key sectors of the economy helped to accelerate
infation in the mid-1960s as the wage-price spiral became more entrenched in the economy.
Table 1. Brazil: Decennial Developments in Capital Flight and
Illicit Financial Flows from Brazil
(in millions of nominal U.S. dollars, in percent, or Gini Coeffcient)
Period
Nature of Crisis
or Economic
Condition
Rate of
Growth of
GDP
(in percent)
Average
Inflation
(in percent)
Central
Govt. Fiscal
Balance
(in percent)
Current
Acc.
Balance
(percent
of GDP)
Income
Inequality
(Gini
Coefficient)
External
Debt
(percent
of GDP)
Broad Capital Flight Illicit Financial Flows
Gross
Outflows
As a
percent
of GDP
Gross
Outflows
As a
percent of
GDP
1960-1969 5.90% 44.16% -3.60% -0.92% 48.76 7.28% 3,979 2.64% 3,097 1.49%
1970-1979
Oil shocks/ high
inflation & growth
7.90% 30.45% 1.96% -4.10% 59.08 23.31% 29,899 2.46% 18,230 1.50%
1980-1989
Hyperinflation/
debt default
3.00% 327.36% 3.21% -1.81% 51.71 37.27% 65,940 2.41% 46,713 1.71%
1990-1999 Hyperinflation 1.70% 843.25% 1.10% -2.11% 52.10 27.28% 159,056 2.56% 85,720 1.38%
2000-2009 3.32% 6.89% -3.28% -0.66% 49.49 24.41% 184,135 1.93% 146,686 1.54%
2010-2012 Stagnation 1.88% 5.69% -2.53% -2.24% 46.51 17.41% 147,223 2.14% 101,162 1.47%
1960-2012 4.52% 252.58% -0.03% -1.68% 52.21 24.39% 590,232 2.20% 401,608 1.50%
Brazil achieved an average rate of growth of 6.0 percent per annum during the 1960s. Economic
growth was mostly driven by the industrial and service sectors rather than by agriculture. In fact,
industrial growth accounted for more than a third of GDP growth during this period, as a signifcant
portion of industrial output drove exports. But prices rose at an average rate of nearly 30 percent
per annum, although there was a deceleration towards the end of the 1960s.
The government resorted to price controls in the 1960s to rein in infation. Controls on a wide range
of industrial goods were implemented through a price council at the Ministry of Finance, while the
Special Secretariat of Supplies and Prices (SEAP) at the Ministry of Planning exercised a general
control over prices. However, price controls also led to increased efforts to circumvent them and to a
proliferation of black markets, as traders tried to make profts by selling only a fraction of their goods
at controlled prices and diverting the bulk through the black market to meet the excess demand.
Widespread price controls and the resulting proliferation of black markets were probably responsible
for the fact that roughly 68 percent of broad capital fight was due to illicit fnancial fows (Table 1).
The factors driving infation in the 1970s were myriad—lax demand management policies, increases
in production costs due to oil price increases, excessive wage increases, and poor agricultural
8 Global Financial Integrity
production. At the beginning of this decade, during 1970-71, infation slowed to an average rate of
around 19.5 percent per annum, decelerating to 15.7 percent in 1972 and to 15.5 percent in 1973;
the frst oil price shock hit Brazil in late 1973. In general, the central government budget was not
a factor driving monetary expansion for much of the 1970s. The favorable scenario had mostly to
do with better revenue performance. Gradually, in the 1970s, open market operations became the
main tool for infuencing aggregate demand management including large open market sales to rein
in liquidity. To the extent that the private sector absorbed the government bonds, this helped policy
makers to curtail the monetization of fscal defcits. Non-infationary sources of fnancing the budget
defcits would, of course, break the link between defcits, the money supply, and prices.
Successive oil price shocks in the 1970s worsened Brazil’s terms of trade and reduced economic
growth. Brazil’s total import bill also increased sharply. For a while, the country managed to ride out
the shock by borrowing cheap petro dollars. But when interest rates rose sharply in the early 1980s
and international commercial lenders curtailed their lending to Latin America, the earlier reliance on
foreign borrowing came back to haunt Brazil. Debt service payments as a share of export earnings
began to rise sharply to a point where it became increasingly diffcult to repay external creditors.
Growth slowed to a crawl, and, in 1987, the government could not make interest payments on its
external debt, which necessitated a rescheduling of public and publicly guaranteed debt.
There was an acceleration of infation during the run up to the debt crisis. More fexible exchange rate
policies coupled with a relaxation of price controls raised the annual rate of infation to 100 percent
per year during 1981-82. Widespread wage-price indexation—supported by an accommodative
monetary policy and the adjustment of key prices—served to fuel infation to 211 percent in 1983,
which increased further to 224 percent in 1984. A comprehensive system of price controls was
re-introduced in April 1985 in an effort to dampen infationary expectations. Infation continued
to be high during the period 1986-89 and economic growth was moderate. There were repeated
attempts to limit infation through a combination of price controls, wage freezes, and modifcations
to the system of price indexation. The objective was to loosen the inertial component of infation.
Because these programs were not supported by appropriate monetary, fscal, and wage policies, the
reduction in infation was short-lived. Once the controls were eased or the indexation re-established,
infation resumed with greater intensity. The rate of infation reached 1,000 percent in 1988.
A combination of frequent devaluations, extensive wage-price and other indexation, and monetary
fnancing of large defcits generated out of control price increases. Hyperinfation peaked at nearly
3,000 percent in 1990. The 1990s started off with declining economic activity accompanied by high
infation—a period that can be characterized as stagfation. Economic growth fell by 1.5 percent a
year on average during 1990-92 while infation averaged 1,040 percent per annum. It was clear by
the early 1990s that achieving stabilization in Brazil required far-reaching economic reform, perhaps
even amendments to the 1988 constitution which had generated a number of fscal problems.
9 Brazil: Capital Flight, Illicit Flows, and Macroeconomic Crises, 1960-2012
An IMF study recognized that the root cause of Brazilian infation had been the monetization of
the public sector fscal defcit, and went on to show that the impact of defcit fnancing through
money creation can become entrenched if economic agents form expectations based on their past
experiences with infation.
11
Moreover, widespread indexation of wages and other contracts not only
created policy inertia but fed infationary expectations themselves.
The Collor I plan, introduced by a new Administration that took offce in March 1990, included a package
of stabilization measures and structural reforms to arrest infation quickly. A signifcant policy action
entailed the “blocking,” or not allowing the monetization, of about two thirds of the fnancial assets in
the economy (M-4) for a period of 18 months. While the measure violated the terms of many fnancial
contracts, it achieved a sharp reduction in liquidity which was complemented by a strengthening
of public fnances. The overall thrust of these policies was to dampen economic activity. Real GDP
growth fell by 1.5 percent per year on average, while annual infation still surged by some 1,040 percent.
But the government made important progress in normalizing relations with external creditors and in
implementing trade liberalization and privatization of many loss-making public enterprises.
A stabilization program designed by Henrique Cardoso called the Plano Real was launched in 1994.
Cardoso would later become the country’s President. Plano Real involved a two-step process. First,
the old currency was replaced by a unit of real value (URV). Then, the Central Bank created a new
currency called the Real which was initially set equal to one U.S. dollar. The URV did away with the
need for indexation because the URV itself was a price index. All existing contracts then had to be
converted based on the URV. It was only then, with a 30-day advance notice by the Bank of Brazil,
that the new currency was introduced. The Real debuted on July 1, 1994, with no surprises and no
price or wage freezes attempted.
Real interest rates rose sharply due to the contractionary impact of the Plan on liquidity. While high
interest rates did rein in infation and attract foreign capital, they led to the deterioration of fscal
accounts due to the asymmetric indexation of expenditures and revenue, which increased nominal
expenditures faster than that of revenue. Moreover, higher real interest rates also led to higher cost
of fnancing the public debt.
The combination of lax fscal policy and tight monetary policy led to an overvaluation of the
currency, which drove capital fight (see the spike in broad capital fight in Chart 3). The exchange
rate peg collapsed in late 1998 after the central bank lost US$14 billion in reserves in two days.
Brazil moved to a de facto foating exchange rate system on January 15, 2000, and the government
introduced the Fiscal Responsibility Law in 2000 in order to control runaway public spending.
Riding on a wave of popular socialist rhetoric, Lula da Silva was elected President in 2002, leading
to investor fears that Brazil may default on its external debt. After all, his Workers Party was quite
radical, and he himself had severely criticized the Plano Real before coming to power. While, on
11.
Dev K. Kar, Government Deficits and Inflation in Brazil: The Experience During 1948-64, IMF Working Paper DM/81/76 (Washington, DC:
International Monetary Fund, October 1981). This working paper explores the impact of monetary and fscal policies on the price level.
10 Global Financial Integrity
balance, Lula maintained macroeconomic stability, there was an increase in gross capital fight in
2004 as a result of sagging investor confdence.
On the whole, this last period covered by our study has been underpinned by much greater
macroeconomic stability, moderately good growth rates, and rising living standards. When the
global economic crisis hit in late 2008, Brazil was better placed to handle the aftermath, but not
without a spike in capital outfows—whether measured as broad capital fight or illicit fnancial fows.
1. Broad Capital Flight and Macroeconomic Crises
We found that, for Brazil, estimates of broad capital fight based on gross outfows were better able to
track macroeconomic crises than net capital fight, gross illicit outfows, or net illicit fows. In general,
outfows through capital fight seem to occur in the aftermath of a crisis (see Chart 3). Thus, the frst
oil shock in late 1973 was followed by a year of signifcant capital fight, which peaked at the end of
1974. Similarly, the second oil shock in 1979 was also followed by more capital fight, which reached a
peak in 1980. In late 1981, there was a spike in interest rates, which resulted in large capital outfows
over the following year, as Brazilian investors began to acquire foreign assets due to large interest rate
differentials in their favor. This was followed by hyperinfation and debt rescheduling in 1986, which
triggered capital fight that spiked in 1987. Hyperinfation continued in the early 1990s, as a result of
which capital fight shot up again in 1993. When the Plano Real was introduced in 1994, it sparked
hope that the stabilization program would stimulate confdence in the economy.
Chart 3. Brazil: Capital Flight and Macroeconomic Crises, 1965-2012
(in millions of nominal U.S. dollars)
0
10,000
20,000
30,000
40,000
50,000
60,000
70,000
1
9
6
5

1
9
6
6

1
9
6
7

1
9
6
8

1
9
6
9

1
9
7
0

1
9
7
1

1
9
7
2

1
9
7
3

1
9
7
4

1
9
7
5

1
9
7
6

1
9
7
7

1
9
7
8

1
9
7
9

1
9
8
0

1
9
8
1

1
9
8
2

1
9
8
3

1
9
8
4

1
9
8
5

1
9
8
6

1
9
8
7

1
9
8
8

1
9
8
9

1
9
9
0

1
9
9
1

1
9
9
2

1
9
9
3

1
9
9
4

1
9
9
5

1
9
9
6

1
9
9
7

1
9
9
8

1
9
9
9

2
0
0
0

2
0
0
1

2
0
0
2

2
0
0
3

2
0
0
4

2
0
0
5

2
0
0
6

2
0
0
7

2
0
0
8

2
0
0
9

2
0
1
0

2
0
1
1

2
0
1
2

M
i
l
l
i
o
n
s

o
f

U
.
S
.

D
o
l
l
a
r
s

Broad Capital Flight
F
o
llo
w
in
g
1
9
7
3
O
il S
h
o
c
k

F
o
llo
w
in
g
1
9
7
9
O
il S
h
o
c
k

In
t
e
r
e
s
t R
a
t
e
S
p
ik
e

P
la
n
o
R
e
a
l
9
/
1
1
A
t
t
a
c
k
s

G
lo
b
a
l F
in
a
n
c
ia
l C
r
is
is

C
o
lla
p
s
e
o
f t
h
e
E
x
c
h
a
n
g
e
R
a
t
e
S
y
s
t
e
m

D
e
b
t C
r
is
is
&
H
y
p
e
r
in

a
t
io
n

11 Brazil: Capital Flight, Illicit Flows, and Macroeconomic Crises, 1960-2012
But the benefcial effect was short-lived and the crawling peg had to be abandoned. The collapse
of the exchange rate system led to massive capital fight in 1998. However, as the program took
hold, the fight of capital was arrested for a few years (1999-2001) when it reached a nadir. The
September 11, 2001, terrorist attacks in the United States led to investor fears and a loss of
confdence in Brazil’s ability to limit the fallout. Capital fight surged in the aftermath of the attacks
and continued through 2003. Over the next three years, 2004-2007, outfows of capital remained
below the peak set in 2003. After that, capital fight spiked up sharply in the run up to the great
recession that began in late 2008.
The sharp jump in capital fight from Brazil in 2012 supports the view that Brazilian investors
decided to pull money out as a result of the European sovereign-debt crisis. Brazil’s Executive
Director to the IMF warned in late 2011 that a worsening of debt problems in the Eurozone countries
could shake investor confdence in Brazil’s fnancial markets, prompting capital fight.
12
Furthermore,
an article published in CNBC in June 2013 expressed the view that Brazil could be most at risk
from capital fight as the country was “…highly vulnerable to currency depreciation and capital
outfows.”
13
According to that article, Morgan Stanley rated Brazil as one of the fve countries most
vulnerable to sudden capital outfows. These views appear to be confrmed by the spike in our
estimates of broad capital fight in 2012.
2. The Nature and Scale of Illicit Financial Flows from Brazil
As noted before, total illicit fnancial fows from Brazil consist of balance of payments leakages
(captured by the HMN measure) and trade misinvoicing (captured by the Gross Excluding Reversals
or GER measure). The GER method estimates outfows of illicit capital through export under-
invoicing and import over-invoicing without netting infows of illicit capital through export over-
invoicing and import under-invoicing. The main reason why we only consider gross outfows of illicit
capital through trade misinvoicing is because the so-called illicit infows represent no beneft to a
country. For instance, import under-invoicing directly results in lower customs duties leading to a
loss of government revenues. A loss in government revenues is not a beneft that should be netted
out from gross outfows of illicit capital.
Outfows due to trade misinvoicing over the 53-year period 1960 to 2012 totaled US$372.3 billion,
while those through balance of payments leakages totaled US$29.4 billion. Barring a slight dip in
the 1990s, total illicit outfows continued to increase signifcantly throughout the decades from an
annual average of US$309.7 million in the 1960s, to US$1.8 billion in the 1970s, which jumped to
$4.7 billion in the 1980s. Illicit outfows increased sharply to an average of US$8.6 billion per year
in the 1990s before ascending to US$14.7 billion in the 2000s. Most of the increase was driven by
trade misinvoicing.
12.
Arnaldo Galvao, “Brazil May Face Capital Flight on European Debt, IMF Director Says,” Bloomberg, 17 October 2011, accessed 16 May
2014, http://www.bloomberg.com/news/2011-10-17/brazil-may-face-capital-fight-on-european-debt-crisis-imf-director-says.html.
13.
Katy Barnato, “This Nation Could Be the Most at Risk From Capital Flight,” CNBC, 14 June 2013, accessed 13 May 2014, http://www.
cnbc.com/id/100815904.
12 Global Financial Integrity
On average, balance of payments leakages account for just 21.2 percent of total illicit outfows,
while the bulk of illicit outfows, 78.8 percent, are related to trade misinvoicing. Broadly speaking,
there seems to be no stability in the way these channels are used to transfer illicit capital. While
the preferred channel has always been through the misinvoicing of trade, its share was nearly 80
percent in the 1960s, growing to 93.5 percent in the 1990s, after which they dropped to 80 percent
in the 1980s and plummeted to just 57 percent in the 1990s. But in the last decade ending 2009, the
share of trade misinvoicing in total illicit outfows increased to 87.4 percent of total outfows.
The sharp increase in the current account defcit in the 1970s relative to the 1960s (see Table 1)
reduced the leakages of both licit and illicit capital from the balance of payments (through greater
use of funds compared to source of funds), triggering an offsetting increase in outfows through
trade misinvoicing. The current account defcit narrowed again over the 1980s, leading to an
increase in the relative importance of balance of payments leakages and a corresponding fall in
trade misinvoicing. However, the current account defcit is not the only factor driving changes in the
relative importance of these two channels for transferring illicit capital. This is because, even as the
current account defcit increased somewhat in the 1990s, HMN-related outfows increased in the
1990s to 57.3 percent of total illicit outfows. For one, there was a sustained decline in regulatory
quality according to the World Bank governance indicators, which could include weaknesses in
customs administration. For another, infation ran at an average annual rate of 843 percent during
the 1990s, which boosted underground economic activities. This in turn seems to have boosted
illicit outfows through trade misinvoicing rather than balance of payments leakages. In our previous
case studies we found a strong link between illicit outfows generated through trade misinvoicing
and the size of the underground economy. In the last decade, the current account defcit narrowed
sharply to just 0.66 percent of GDP, which reduced outfows through the balance of payments and
increased the use of trade misinvoicing to 87.4 percent of total outfows.
Export under-invoicing is the primary mechanism by which Brazilian traders misinvoice trade to shift
capital abroad illicitly. Over the period 1960-2012, some 73.7 percent of trade-related illicit outfows
occurred through export under-invoicing. Import over-invoicing accounted for just 26.3 percent of
total trade misinvoicing. The imposition of state and other taxes, such as social taxes, on imports,
on top of the tax that goes to the central government, may raise the total import taxes to such a
level that it is no longer advantageous for Brazilian companies and traders to over-invoice imports—
particularly in relation to the effective corporate tax rate, which has hovered around 24 percent in
recent years according to the accounting frm PricewaterhouseCoopers.
14
Companies typically
do not gain by paying a higher import cost through over-invoicing when they cannot offset it by
paying lower corporate taxes. In short, as long as the marginal import duty is higher than marginal
corporate tax rate, there is no gain in shifting the higher import costs on to corporate taxes. Hence
14.
PwC, “Brazil,” in Worldwide Tax Summaries: Corporate Taxes 2013/14 (New York: PwC, 2013), 265-6, accessed 14 July 2014, http://
www.pwc.com/gx/en/tax/corporate-tax/worldwide-tax-summaries/assets/pwc-worldwide-tax-summaries-corporate-2013-14.pdf.
13 Brazil: Capital Flight, Illicit Flows, and Macroeconomic Crises, 1960-2012
the preferred method has been to under-invoice exports. While export under-invoicing was the
predominant form of trade misinvoicing in the 1960s, 1980s, 1990s, and the 2000s, import over-
invoicing dominated most of the 1970s. More research into why the pattern of trade misinvoicing
fipped in the 1970s might be carried out, but this is a matter that is outside the scope of the present
study.
Chart 4. Brazil: Illicit Financial Outflows and the Underground Economy, 1960-2009
15

(percent of GDP)
0.60%
0.80%
1.00%
1.20%
1.40%
1.60%
1.80%
0.00%
10.00%
20.00%
30.00%
40.00%
50.00%
60.00%
1960-1969 1970-1979 1980-1989 1990-1999 2000-2009
P
e
r
c
e
n
t

o
f

G
D
P

P
e
r
c
e
n
t

o
f

G
D
P

Average Underground Economy to GDP (leD axis) Average Illicit Financial Flows to GDP (right axis)
Average Illicit Financial Outflows to GDP (Right Axis) Average Underground Economy to GDP (Left Axis)

Chart 4 shows that illicit outfows have tended to follow rather closely the share of the underground
economy to GDP ratio.
16
The increase in outfows to GDP in the most recent decade ending 2009
is an exception to this pattern. Because the underground economy (as a share of offcial GDP) is a
proxy for the overall state of governance, it is not surprising to fnd close association between the
cross-border transfer of illicit capital and the underground economy as depicted in Chart 4.

15.
Estimates of the underground economy were obtained through the monetary approach (see Appendix II for details on methodology).
16.
See Appendix II
14 Global Financial Integrity
15 Brazil: Capital Flight, Illicit Flows, and Macroeconomic Crises, 1960-2012
III. A Model of Capital Flight and
Illicit Financial Flows from Brazil
We develop a structural equations model (SEM) to examine the drivers and dynamics of both illicit
fnancial fows and capital fight from Brazil. In other words, we model gross outfows of licit and
illicit capital as well as outfows that are purely illicit. Inward capital transfers are not netted out from
such outfows.
This is a larger model than the one we developed in the case of our earlier studies of illicit fows out
of India, Mexico, the Philippines, or Russia. It is larger because (i) unlike in other case studies, the
present SEM seeks to explain nominal income (GDP) endogenously; (ii) capital formation, which is a
key factor driving offcial GDP, is also endogenous; and, (iii) illicit fows and capital fight are shown
to be driving each other either directly or indirectly via their impact on the underground economy.
There are nine structural equations and one behavioral equation specifying how infationary
expectations are formed. Six of the nine structural equations relate to the offcial economy, i.e.,
government expenditures, government revenues, broad money supply, formation of prices as a
result of the interaction between monetary and fscal policies, gross fxed capital formation (relating
to both the offcial and private sector), and nominal income. Three other equations capture how
broad capital fight, illicit fows, and the underground economy interact with the offcial economy.
Before estimating the model, we address the issue of identifcation of the structural equations. If
any equation is under-identifed, then the parameters of the equation cannot be estimated, so that
the entire model cannot be simulated. It must be possible for numerical estimates of the structural
equation to be obtained from the estimated reduced-form coeffcients, so we need to impose
the order condition for identifcation for each equation. The order condition, which is a necessary
condition for identifcation, states that the number of predetermined variables excluded from the
equation must not be less than the number of endogenous variables included in that equation less
one. We can see that, in fact, each structural equation is over-identifed.
Researchers have widely used two methods for estimating an interdependent system of structural
equations—the three-stage and two-stage least squares methods (3SLS and 2SLS, respectively).
While both 3SLS and 2SLS provide consistent estimates, we use the 2SLS technique mainly
because there is no gain in asymptotic effciency in small samples. The benefts of applying the
3SLS cannot be realized in a sample size of some 60 observations.
The next several pages will explain how the various sectors of the model are derived.
16 Global Financial Integrity
i. Government Sector
Apart from the fact that the government plays an important part in all developing countries, the
exact specifcation of how central government revenues and expenditures behave is necessary
to capture their interactions with the monetary sector and the resulting formation of price
developments and expectations. Note that we focus only on the central government rather than
the general government accounts, which consolidate the central government with state and local
governments. The main reason for the narrower focus on the central government is due to the fact
that data on consolidated general government are not available for the period 1965-2011.
The model makes the assumption that the government’s desired amount of real expenditures
depends on the prevailing level of real income—that is, the government strives to at least
maintain the real value of its expenditure outlays because failing to do so would lead to economic
contraction. While the Brazilian government has cut back on expenditures, such cuts have seldom
been in real terms. In general, this is a reasonable assumption. Hence, in logarithms (ln), the
relationship is:
ln (G/P)
D
= a
0
+ a
1
ln Y
t
, a
1
>0
where G is nominal government expenditures, P the price level (measured by the consumer price
index), and Y is real income (GDP). Actual current real expenditures are assumed to adjust current
desired expenditures and actual real expenditures in the previous period, that is:
ln (G/P)
t
= α [ln (G/P)
D
– ln (G/P)
t-1
], 1 > α >0, where α is the adjustment coeffcient. We
eliminate desired real expenditures through substitution which yields:
ln (G/P)t = αa
0
+ αa
1
ln Y
t
+ (1- α) ln (G/P)
t-1

or, ln G
t
= αa
0
+ αa
1
ln Y
t
+ (1- α) ln (G/P)
t-1
+ ln P
t
, 1 > α >0
The reduced form equation for government revenue is formulated similarly. Thus:
ln R
t
= βb
0
+ βb
1
(ln Y
t
+ ln P
t
) + (1-β) ln R
t-1
, 1 > β >0
ii. Money Supply Process
We specify the money supply process according to the Brunner-Meltzer (BM) formulation, in that the
nominal money supply is a function of the monetary base, the ratio of currency to demand deposits,
and the discount rate. The money supply is postulated to vary positively with the monetary base,
which is the amount of money issued by the central bank, negatively with the amount of currency
relative to demand deposits, and positively with the discount rate. The drawback of the Brunner-
17 Brazil: Capital Flight, Illicit Flows, and Macroeconomic Crises, 1960-2012
Meltzer formulation is that government revenues and government expenditures do not enter the
equation endogenously. The quantity theory of money, on the other hand, explicitly incorporates the
impact of fscal policy on the money supply process, but the equation is an identity. The Brunner
Meltzer formulation is:
ln M
t
= c
0
+ c
1
ln (MB
t
) + c
2
ln IR
t
+ c
3
ln CR
t
iii. Formation of Prices
The equation for the price level (P) is derived from a standard function for real money demand. We
assume that in a developing country like Brazil, the expected rate of infation, rather than the rates
of return on fnancial assets, is more likely to refect the true opportunity cost of holding real money
balances. We also assume that the actual stock of real money adjusts by a constant proportion to
the difference between the real money demand and the stock of money in the previous period:
∆ln (M/P)
t-1
= γ [ln (M/P)
D
– ln (M/P)
t-1
], 1 > γ > 0,
where γ is the adjustment coeffcient. The demand for real money balances in developing countries
is therefore formulated as:
ln (M/P)
D
t
= d
0
+ d
1
ln Y
t
– d
2

t
d
1
, d
2
> 0
where ∏
t
, the expected rate of infation, serves as a proxy for the opportunity cost for holding
money in an economy with administered interest rates at least over a signifcant period of time. The
demand for real money balances is eliminated through substitution yielding:
ln (M/P)
t
= γ d
0
+ γ d
1
ln Y
t
- γ d
2

t
+ (1 - γ) ln (M/P)
t-1

or, ln P
t
= - γ d
0
- γ d
1
ln Y
t
+ γ d
2

t
- (1 - γ) ln (M/P)
t-1
+ ln M
t
, 1 > γ >0
iv. Real Sector
The real sector in this model consists of two structural equations explaining nominal GDP and gross
fxed capital formation. Nominal GDP is specifed as a standard Cobb-Douglas production function,
which links inputs and outputs. These links have been tested for many countries over several
decades. The formula is:
GDP = P f(K, L)
which specifes that the total goods and services produced in an economy (GDP) depends
on productivity (P), which is popularly known as technology and is also a function of capital
18 Global Financial Integrity
investments and labor input. Because the share of capital and labor, Ѳ, will vary by country, the
testable production function is:
GDP = PKѲL
1-Ѳ

ln GDP
t
= e
0
+ e
1
Ѳ ln K + e
2
(1-Ѳ) ln L
where ln P is the constant in the regression. The coeffcients of capital and labor sum to one or
come very close to it.
The second equation for gross fxed capital formation serves to link the impact of the underground
economy (or the overall state of governance) on the offcial economy. Hence, the impact of
governance issues on the economy had to be modeled indirectly because the Cobb-Douglas
production function does not allow the inclusion of other factors that could impact production.
The investment function is also specifed as a function of normal factors like interest rate, nominal
income (GDP), investment in the previous period, external debt, and the underground economy. The
interest rate is proxied by the expected rate of infation because, in an economy where rates have
been administratively fxed for some time or where rates do not fully refect the supply and demand
for loan-able funds, the expected rate of infation serves as the opportunity cost of holding money.
Nominal income is a standard explanatory variable in investment functions, while investment in the
current period has often been found to be signifcantly linked to investment in the previous period in
other countries. Investment may also be driven by contracting external debt, so we test whether this
is true for the period as a whole. The underground economy may adversely impact growth of the
offcial economy if the overall productivity of capital invested there is lower than their rate of return
in the offcial economy. This would be true if, for example, capital is invested in carrying out black
market transactions, to fnance contraband, or to invest in speculative real estate. The gains from
such investments accrue to the corrupt, with the benefts accruing to the larger economy.
ln K
t
= δf
0
+ δf
1
ln K
t-1
+ δf
2
ln UE
t
+ δf
3
ln GDP
t
+δf
4
In ExtDebt
t
+ ∏
t

The fnal equation of the offcial economy that closes the loop is expected infation, ∏
t
. Infationary
expectations are formulated along similar lines as Cagan, where such expectations are formed
according to an adaptive process. This means that an increase in actual infation translates into an
increase in infationary expectations. The equation is as follows:

t
= ξ∆ln P
t
+ (1 - ξ) ∏
t-1
, 1 > ξ > 0
where ξ denotes the coeffcient of expectations and ln P
t
is the current rate of infation. If all
economic agents were somehow able to formulate the infationary expectations
17
with perfect
foresight, past experiences with infation would play no role and the adjustment parameter would
17.
Dev Kar, “The Brazilian Financial Sector: An Empirical Study in Evolution” (PhD Dissertation, George Washington University, 1982).
19 Brazil: Capital Flight, Illicit Flows, and Macroeconomic Crises, 1960-2012
be equal to one. In the absence of perfect information regarding future outcomes of infation,
however, the Cagan specifcation allows infation in prior periods to play a role in current infationary
expectations. The adjustment coeffcient will vary depending upon the expectations of economic
agents in an infationary environment. If, for example, as in Brazil, a country experiences high and
highly variable infation or hyperinfation, the expectations of economic agents will be skewed
towards assigning more weight to infation in the current period. In other words, the infationary
expectations of economic agents would be driven mainly by their experience with infation in
the current period (i.e., the adjustment coeffcient will be relatively high, say 0.9 or higher) rather
than their past experience of infation. We use different adjustment coeffcients to maximize the
signifcance of infationary expectations wherever the variable is used in the model.
v. Underground Economy, Illicit Flows, and Capital Flight
The underground economy is independently estimated using the monetary approach prior to
its inclusion in the model (see Appendix II for methodology). Within the model, the underground
economy is cast as a function of just three variables—infation, illicit outfows, and growth of the
offcial economy.
ln UE
t
= ψg
0
+ ψg
1
ln P
t
+ ψg
2
ln IFF
t
+ ψg
3

t
where Ẏ
t
represents economic growth. Infation, and particularly hyperinfation, severely impacts
those on fxed income and forces them to fnd alternative sources of income in the informal and
underground economies. The informal economy is not necessarily illegal (such as retail trade).
But the informal economy is linked to the underground economy by the fact that workers in both
generally do not pay taxes. Illicit outfows lead holders of such assets to not report income derived
from the return on those assets which in turn adds to the underground economy. We would expect
growth in the offcial economy to be negatively related to the underground economy because higher
growth rates would tend to divert resources to the offcial economy, as more labor and capital seek
legitimate opportunities rather than undertake the risks associated with illegal activities.
We introduce equations for both broad capital fight and illicit fows. In this way, we can study the
different factors that drive them. The hypothesis is that broad capital fight, as estimated by the
World Bank Residual approach adjusted for trade misinvoicing, is driven by both macroeconomic
and governance-related factors. Income inequality, which could be a structural factor driving
capital fight, could not be included because of the paucity of data on the Gini.
Given data limitations, we model capital fight as follows:
ln CapFlight
t
= λh
0
+ λh
1
ln IFF
t
+ λh
2
ln P
t
– λh
3
ln GDP
t
– λh
4

t

20 Global Financial Integrity
The above formulation uses illicit fows as a proxy for the state of overall governance, rather than
the underground economy, because we want to capture interactions between the two. Given that
both capital fight and illicit fows are adjusted by trade misinvoicing, we are basically assessing
the signifcance of illicit fows (as captured by the net errors and omissions or HMN) in explaining
the gap between the source and use of funds underlying the World Bank Residual approach. We
would expect illicit fows to be positively related to broad capital fight. Price developments are also
likely to drive capital fight to the extent that rising prices erode the real value of domestic assets,
driving investors out in search of more stability, while rising nominal income and real growth are
expected to stem capital fight as investors gain more confdence in the domestic economy and the
attractiveness of domestic over foreign assets increases. We found that macroeconomic factors,
such as prices, nominal income, and economic growth, were not signifcant in explaining illicit fows.
Therefore, illicit fows are formulated simply as a function of the underground economy, that is:
ln IFF = μ
0
i
0
+ μi
1
ln UE
t

vi. The Complete Model
The complete ten-equation SEM that is simulated is as follows:
ln G
t
= αa
0
+ αa
1
ln Y
t
+ (1 - α) ln (G/P)
t-1
+ α ln P
t
ln R
t
= βb
0
+ βb
1
ln GDP
t
+ (1 - β) ln R
t-1
ln M
t
= c
0
+ c
1
ln MB
t
+ c
2
ln IR
t
+ c
3
ln CR
t
ln P
t
= - γ d
0
- γ d
1
ln Y
t
+ γ d
2

t
- (1 - γ) ln (M/P)
t-1
+ ln M
t
ln GDP
t
= e
0
+ e
1
Ѳ ln K + e
2
(1 - Ѳ) ln L
ln K
t
= δf
0
– δf
1
ln UE
t
+ δf
2
ln CapForm
t-1
+ δf
3
ln GDP
t
+ δf
4
ln ExtDebt
t
- δf
5

t


t
= ξ∆ln P + (1 - ξ) ∏
t-1
ln UE
t
= ψg
0
+ ψg
1
ln P
t
+ ψg
2
ln IFF
t
- ψg
3

t
ln CapFlight
t
= λh
0
+ λh
1
ln IFF
t
+ λh
2
ln P
t
– λh
3
ln GDP
t
– λh
4

t
ln IFF = μ
0
i
0
+ μi
1
ln UE
t
To sum up, the endogenous variables determined within the SEM comprise the following: G and R
are the nominal expenditures and revenues of the central government respectively, M the supply of
broad money, P the price level as captured by the consumer price index, GDP the nominal income,
K the gross fxed capital formation consisting of both public and private investment, ∏ the expected
rate of infation, UE the underground economy, CapFlight is broad capital fight as estimated by the
World Bank Residual model adjusted for trade misinvoicing, and IFF represents illicit fnancial fows
as estimated by the Hot Money Narrow (HMN) method based on net errors and omission of the
balance of payments adjusted for trade misinvoicing. Both CapFlight and IFF estimates are based on
outfows only; infows are not netted out from outfows. The rationale for focusing only on outfows is
that, because fows are illicit in both directions (as a signifcant portion of broad capital fight is also
illicit), it makes little sense to net out such fows, which would be akin to the concept of net crime.
21 Brazil: Capital Flight, Illicit Flows, and Macroeconomic Crises, 1960-2012
The exogenous variables in the above SEM are real income Y, monetary base created by the
government MB, discount rate of interest IR, the currency to demand deposit ratio CR, labor supply
L, level of outstanding external debt ExtDebt, real economic growth Ẏ
t
plus all lagged variables.
vii. Results of Dynamic Simulation of the SEM
The main fndings of the SEM on Brazil are as follows:
1. The model results show that the underground economy is the main link through which illicit
fows impact the Brazilian economy at large and are in turn impacted by developments in the
broader economy. The two-way interactions of illicit fows and the broader economy are not
direct but indirect. For example, illicit fows were found to be a signifcant driver of Brazil’s
underground economy, which in turn negatively impacted capital formation. In other words, as
the underground economy grew, it diverted resources away from the offcial economy, leading
to lower capital formation.
2. Capital formation, in turn, is positively and signifcantly related to economic growth. Hence, to
the extent that the underground economy acts as a drag on investment in the offcial economy,
illicit outfows lower the potential rate of growth (defned as the rate of growth without illicit
outfows). Hence, illicit outfows represent a signifcant loss to the Brazilian economy.
3. Model results also show that the underground economy itself drives illicit outfows—the larger
the underground economy, the greater the capacity to generate illicit outfows. However, apart
from illicit outfows, we did not fnd infation or real economic growth to be signifcant drivers of
the underground economy.
4. Illicit fows are signifcantly related to capital fight. A 1.0 percent increase in illicit outfows leads
to a 0.83 percent increase in capital fight.
5. Government revenues are mainly driven by nominal income (GDP). In contrast, lagged revenues
were not signifcant in explaining the current period’s revenue collections. In contrast, the
previous period’s real expenditures were signifcant in determining current expenditures. In
spite of high and highly variable infation as well as hyperinfation, we fnd that, in general, the
Government did not allow expenditures to decline in infation-adjusted terms. That is not to say
that real expenditures were not cut as part of fscal adjustment over a specifc period, but—for
the period as a whole—that has certainly not been the case.
6. Prices are mainly driven by increases in broad money. Real GDP had the expected negative
sign—in other words real economic growth is negatively related to growth, although the
signifcance is only at the 90 percent confdence level. It was surprising to fnd that infationary
expectations did not feed back into prices in a signifcant manner, although there is a positive
association. Perhaps the adaptive error learning process does not adequately capture the
formation of expectations when infation is highly variable and there are episodes of hyperinfation.
As expected, the real money stock in the previous period was statistically signifcant and
negatively related to prices in the current period. Except as noted, the signs of the variables and
their statistical signifcance are consistent with those predicted by monetary theory.
22 Global Financial Integrity
7. Broad money supply was formulated according to the Brunner-Meltzer theory. The monetary base
and the currency to demand deposit ratio were strongly signifcant and were found to drive the money
supply. The discount rate was signifcant only at the 90 percent confdence interval, and the coeffcient
was much smaller than either base money or the currency ratio. We found no evidence that over the
period 1965-2011, Brazil’s fscal policy played any role in driving infation. There are two main reasons
behind this fnding. First, the central government fscal balance remained in surplus (i.e., revenues
exceeded expenditures) for two continuous extended periods, 1970-1985 and 1987-1995. Hence, during
these extended periods, monetary policy variables are not impacted by fscal issues but by money
market equilibrium. The periods 1970-1985 and 1987-1995 can be called “monetary dominant”. In
contrast, monetary dominance was interspersed by two continuous periods of signifcant fscal defcits,
1960-1969 and 1996-2012. These periods are said to be fscally dominant in that monetary policy is
typically subordinated either through direct fnancing in the form of central bank credits and money
creation or through domestic bond fnancing. The latter tends to crowd out private investment as
interest rates rise. Regardless of the fact that fnancing can also take place through some combination
of monetary expansion, domestic bond sales, and foreign fnancing, the fact remains that large defcits
tend to impose a fscally dominant regime.


ln G
t
= -11.894 + 0.292 ln Y
t
+ 1.265 ln (G/P)
t-1
+ 0.955 ln P
t

[-1.28] [0.71] [3.26]*** [29.43]*** R
2
= 0.999 SE = 0.401
ln R
t
= 13.417 + 0.939 ln GDP
t
+ 0.103 ln R
t-1

[13.97]*** [13.15]*** [1.52] R
2
= 0.999 SE = 0.483
ln M
t
= 1.271 + 0.996 ln MB
t
+ 0.075 ln IR
t
+ 1.197 ln CR
t

[4.08]*** [85.14]*** [1.792]* [6.70]*** R
2
= 0.999 SE = 0.321
ln P
t
= -4.057 – 0.385 ln Y
t
+ 0.015 ∏
t
- 0.670 ln (M/P)
t-1
+ 0.982 ln M
t

[-1.33] [-1.82]* [1.57] [-4.89]*** [93.91]*** R
2
= 0.999 SE = 0.301
ln GDP
t
= -13.220 + 0.248 ln K + 0.740 ln L
[-8.05]***[2.94]*** [8.97]*** R
2
= 0.999 SE = 0.072
ln K
t
= -1.214 + 0.147 ln K
t-1
− 0.321 ln UE
t
+ 1.060 ln GDP
t
+ 0.102 In ExtDebt
t
+ 0.015 ∏
t

[-5.25]*** [1.66] [-2.06]** [11.32]*** [1.30] [1.63]

t
= 0.9ln P +0.1 ∏
t-1
R
2
= 0.999 SE = 0.133
ln UE
t
= 0.873 – 0.118 ln P
t
+ 1.078 ln IFF
t
– 0.025Ẏ
t

[0.11] [-0.38] [3.67]*** [-0.01] R
2
= 0.999 SE = 0.628
ln CapFlight
t
= 7.913 + 0.828 ln IFF
t
+ 0.274 ln P
t
– 0.098 ln GDP
t
– 1.932Ẏ
t

[1.17] [2.95]*** [1.295] [-0.242] [-1.047] R
2
= 0.998 SE = 0.468
ln IFF = -3.854 + 1.029 ln UE
t

[-12.55]*** [66.54]*** R
2
= 0.998 SE = 0.588
Structural and Behavioral Equation Estimates
23 Brazil: Capital Flight, Illicit Flows, and Macroeconomic Crises, 1960-2012
The shift of the policy stance from one of fscal to monetary dominance only to relapse into the
former regime is the main reason why we fnd no evidence that Brazil’s fscal policy over the
period as a whole played any signifcant role in driving infation. That does not mean we can rule
out the monetary impact of large fscal imbalances in sub-periods, such as 1996-2012. Rather,
that impact would also depend on whether defcits were primarily fnanced through monetary
expansion.
8. Available evidence based on IMF Country Reports and Staff Reports for Article IV Consultations
show that, while defcits were mainly fnanced through central bank credits and monetary
expansion during much of the earlier period 1960-1969, bond fnancing together with foreign
fnancing became much more important in the more recent period.
18
This is another reason why
researchers are unlikely to fnd any signifcant link between fscal defcits, the money supply,
and infation. This is quite a different scenario from the earlier period, 1948-1964, in Brazil when
there was a strong link between defcits, money supply, and infation. This led to an asymmetric
response of revenue and expenditures to infation (due to the faster speed of adjustment of
expenditures than revenues to infation) which further widened the defcits leading to more
money creation and infation in a vicious circle.
19

9. Nominal income (GDP), which was formulated as a standard Cobb-Douglas function, is driven
by capital formation (gross public and private investment) and labor supply. Productivity and
technology are assumed to remain fxed. Both capital and labor were found to be signifcant at
the 95 percent confdence interval (with their coeffcients adding to one).
10. Nominal income was found to be a signifcant driver of gross fxed investment. While
contracting new external debt seemed to have a positive impact on capital formation, the
relationship was not signifcant at the 90 percent level. The interest rate (based on the expected
rate of infation as an opportunity cost of holding money) was also not signifcant in explaining
investment, perhaps due to the fact that interest rates were administratively set for many years
in Brazil under successive governments.
To summarize, each component of the SEM can be classifed into three broad categories—
macroeconomic, behavioral, and target (or what we are trying to explain). How these broad classes
of drivers impact the target variables (e.g., underground economy, capital fight, and illicit fnancial
fows) are captured in Table 2.
18.
International Monetary Fund, “Fiscal Sustainability and Monetary Versus Fiscal Dominance: Evidence from Brazil, 1991-00,” in Brazil:
Selected Issues and Statistical Appendix, IMF Country Report No. 01/10 (Washington, DC: IMF, January 2001), 9; International Monetary
Fund, “Brazil: Staff Report for the 2012 Article IV Consultation,” in Brazil: 2012 Article IV Consultation – Staff Report; Public Information
Notice on the Executive Board Discussion; and Statement by the Executive Director for Brazil, IMF Country Report No. 12/191
(Washington, DC: IMF, July 2012), 22, 57, 73.
19.
Kar, “Government Defcits and Infation.”
24 Global Financial Integrity
Table 2. Brazil: Components of the Structural Equations Model and
How They Impact Target Variables

Nature of the Equation in the Model
Direct or Indirect Impact
Between and Among Targets
and Intermediaries
Significance of Direct and Indirect
Impact
Number of Times
Variable Appears
within Model
Macroeconomic
Government Expenditures (G) Direct P on G Significant One
Government Revenues (R ) Direct GDP on R Significant One
Broad Money Supply (M) Indirect through prices Significant Two
Formation of Prices (P) Direct M, π on P Significant; insignificant Five
Gross Capital Formation (K) Direct GDP, UE, π on K Both significant; insignificant Two
Nominal Income (GDP) Direct K on GDP Significant Four
Behavioral
Inflationary Expectations (π) Direct P on π Significant Three
Target Variables
Underground Economy (UE) Direct P, IFF on UE Insignificant; significant Three
Capital Flight (CapFlight, CF) Direct IFF, P, GDP on CF Only IFF significant One
Illicit Financial Flows (IFFs) Direct UE on IFF Significant Three

Impact of exogenous variables (i.e., those determined outside the SEM) are not shown in the table.


The following observations are salient:
• Of the ten equations listed in the left-most column, six are macroeconomic, one is behavioral,
and three are target variables in the sense that they are of particular interest to the study.
We note that only prices (P) and GDP have a direct impact on the target variables such as
the underground economy and capital fight. Although prices have a direct impact on the
underground economy and capital fight, the impact was found to be statistically insignifcant.
Even nominal income (GDP) was not found to be statistically signifcant in explaining capital fight.
That is why macroeconomists fnd it diffcult to trace the direct impact of economic variables on
the underground economy (except for taxes, which are not modeled in this study). It is therefore
not surprising that academic literature has only found a weak link between macroeconomic
variables (e.g., fscal defcits, interest rates, and infation) and capital fight. There are hardly any
empirical studies on illicit fnancial fows thus far.
• Macroeconomic variables tend to interact within each other rather than with the target variables,
such as illicit fows and the underground economy.
• However, macroeconomic and behavioral variables interact with the underground economy,
capital fight, and illicit fows in complex ways. For instance, the underground economy is found
to have a signifcant negative impact on offcial investment. It seems that a faster rate of growth of
the underground economy can deplete investment that would otherwise be invested in the offcial
economy. However, the relationship is only signifcant at the 90 percent level. Investment, in turn,
drives economic growth. Hence, the underground economy, by hampering legitimate investment,
indirectly deters growth, although the direct (negative) relationship between the two was not
found to be statistically signifcant.
• The underground economy tends to be driven mainly by other governance-related drivers, such
as illicit fows, rather than by macroeconomic or behavioral factors.
• Prices and nominal income variables percolate the most throughout the ten-equation system (fve
25 Brazil: Capital Flight, Illicit Flows, and Macroeconomic Crises, 1960-2012
and four times respectively) followed by infationary expectations, the underground economy, and
illicit fows (thrice each). The more endogenous variables appear in the system, the higher the risk
that small deviations between actual and simulated values can get compounded.
viii. Inequality, Capital Flight, and Illicit Flows
The following results from a multiple regression analysis seek to throw some light on whether there
is a link between economic growth, income inequality, and capital fight.
Table 3. Multiple Regression Analysis: Links between Economic Growth,
Income Inequality, and Capital Flight
Source SS df MS
Number of obs = 42
F (3,38) = 1937.31
Prob > F = 0.0000
R-squared = 0.9935
Adj R-squared = 0.9930
Root MSE = 1.0465
Model 6365.00073 3 2121.66691
Residual 41.6161147 38 1.09516091
Total 6406.61685 41 156.258947
ced_ger_In Coef. Std. Err. t P>|t| [95% Conf. Interval]
gdp_growth 7.795613 4.274226 1.82 0.076 -0.8571048 16.44833
Gini 0.961393 0.0522824 1.84 0.074 -0.009701 0.2019795
gdp_In 1.004467 0.0178712 56.21 0.000 0.9682884 1.040645
_cons -13.38388 3.015685 -4.44 0.000 -19.48882 -7.278946
One of the main impediments to including the Gini into the SEM is the fact that continuous data
on the Gini is only available for 1976 to 2011, which will introduce small sample bias in dynamic
simulations. So we interpolated the series for the period 1970-1975 before regressing broad capital
fight on the factors shown in the table above. The regression results show that worsening income
inequality also seems to drive capital fight, although the relationship is only signifcant at the 90
percent level.

26 Global Financial Integrity
27 Brazil: Capital Flight, Illicit Flows, and Macroeconomic Crises, 1960-2012
IV. The Legal and Policy Environment in Brazil
Our fnding that illicit fows through trade misinvoicing comprise the largest proportion of capital
fight from Brazil suggests curbing capital fight will require strong customs and tax enforcement
and oversight. Brazil has also long struggled with corruption, a problem that led to some public
unrest preceding the 2014 FIFA World Cup, but our fnding of the persistent size of Brazil’s
underground economy suggests that the country faces much broader governance issues. Brazil has
made great strides in recent years towards bringing its anti-money laundering regime in line with the
international standards embodied in the Financial Action Task Force (FATF) Recommendations, but
these legal changes have not necessarily been accompanied by effective enforcement.
i. Customs, Trade, and Tax
Responsibility for customs enforcement lies with the Secretariat of the Federal Revenue of Brazil
(RFB). Brazil has taken strong steps to curb trade fraud through imports recently, establishing the
National Centre for Customs Risk Management in 2012 to coordinate analysis and investigations of
fraudulently documented transactions.
20

Traders must declare all imports and exports of goods through the Integrated Foreign Trade
System (SISCOMEX), a computerized tracking system operated by RFB. Imports are processed
according to a risk-based system similar to that in place in many other countries. In recent years,
12-16 percent of imports were subject to inspection, only half of which were physically inspected.
Risk assessment is based entirely on the contents of the import declaration as fled; additional
documentation is requested only if the shipment is fagged for further inspection.
Brazil is a signatory to the WTO Customs Valuation Agreement, under which Brazil agrees to apply
the “transaction value” principle, valuing imported or exported goods at the “price actually paid
or payable” for the goods—in effect, the price that is refected on the invoice between parties. The
agreement allows for customs authorities to request additional documentation to support the stated
price in cases where it is deemed suspect, but in practice, Brazil accepts the declared value in 99.8
percent of all transactions.
21

While Brazil’s customs regime appears suffciently rigorous for a country its size, its apparent
shortcomings given the large volume of illicit outfows through trade is not surprising. Customs
enforcement in Brazil, as in most other countries, is intended to ensure the collection of proper tax
and tariff revenue, and Brazil does not tax exports (except for a few select goods). Thus there is
no revenue to be gained from scrutinizing export transactions. However, nearly three-quarters of
Brazil’s illicit outfows through trade occur via exports.
20.
World Trade Organization, Trade Policy Review: Brazil, WT/TPR/S/283 (Geneva: WTO, 17 May 2013), 47, Accessed 14 July 2014, http://
www.wto.org/english/tratop_e/tpr_e/tp383_e.htm.
21.
Ibid., 48.
28 Global Financial Integrity
Instead, export transactions ultimately fall primarily under the purview of tax enforcement, as the
value received for exported goods strongly affects the rate of income tax the exporting company
will pay. Specifcally, companies will under-invoice exports in order to reduce the amount of proft
they declare in Brazil, generally under a tacit agreement with the importer to remit the remaining
value to an offshore account controlled by the company’s owner.
Brazil’s efforts to address the problems of abusive transfer pricing are admirable and worth noting.
Transfer pricing is the method by which related companies (i.e., companies with some common
ownership) account for the movement of goods and services between jurisdictions. In general, it
is governed by the international “arm’s-length principle,” which purportedly mimics the accounting
of similar transactions between unrelated parties. In reality, though, transfer pricing rules can be
readily manipulated to minimize taxable profts in high-tax jurisdictions and shift capital to low-tax
jurisdictions. Although for many countries much of the capital fight that occurs through abusive
transfer pricing is related to services and intangibles and thus not captured in our data, as Brazil
has demonstrated, the tactics to address abusive transfer pricing and trade misinvoicing are very
similar.
Brazil has moved away from the arm’s-length principle by instituting more objective methods for
determining an appropriate price, establishing ceilings for deductible expenses and fxed proft
margins on certain transactions.
22
Brazil also recently required imports and exports of certain
commodities to be priced in accordance with current world market prices, independent of a
company’s costs or the structure of the transaction.
23
Brazil also recently instituted a law extending
this regime in order to subject transactions with entities located in tax havens to the same strict
scrutiny as transactions with related parties.
24
This is a simple but powerful tool to address the large
role tax havens play in trade misinvoicing, but it is too early to statistically tell whether it has been
effective.
ii. Transparency and Governance
Brazil has consistently fared poorly in common indices of governance and corruption. While much
of the legal framework needed to combat corruption has been put in place over time, its success
depends on the amount of political will available to strictly and fervently enforce it over the medium-
and long-term.
25

22.
PwC, “Brazil,” in International Transfer Pricing 2013/14 (New York: PwC, 2013), 284, accessed 3 July 2014, http://www.pwc.com/gx/en/
international-transfer-pricing/assets/itp-2013-fnal.pdf.
23.
Law 12,715/12, effective January 1, 2013. See also PwC, “Brazilian Federal Revenue Department issues further guidance on changed
introduced by Law 12715,” Tax Insights, January 9, 2013, http://www.pwc.com/en_GX/gx/tax/ newsletters/pricing-knowledge-network/
assets/pwc-brazil-law12715-guidance.pdf.
24.
RFB Normative Instruction No. 1037, June 4, 2010. See also Walter Stuber and Adriana Maria Gödel Stuber, “Brazil: Brazilian Tax
Authorities Issue a New List of Favored Taxation Countries,” Mondaq, June 10, 2010, accessed 3 July 2014, http://www.mondaq.com/
x/102618/Income+Tax/Brazilian+Tax+Authorities+Issue+a+New+List+of+Favored+Taxation+Countries.
25.
“Corruption by Country: Brazil,” Transparency International, accessed 14 July, 2014. http://www.transparency.org/country#BRA.
29 Brazil: Capital Flight, Illicit Flows, and Macroeconomic Crises, 1960-2012
The use of anonymous legal entities is a signifcant transparency issue facing many countries
in both the developing and developed world, and Brazil is no exception. While Brazil operates a
central registry of legal entities and has made this registry open to the public, this registry does not
collect information on the ultimate “benefcial owner” of legal entities controlled by foreign citizens
or foreign legal entities, suggesting that it is relatively easy to form and operate an anonymous
company in Brazil. Furthermore, Brazil does not regulate corporate service providers, who in many
other countries serve as gatekeepers, collecting identifcation information for incorporators and
screening for money laundering risks.
Brazil has effectively instituted transparency surrounding the licensing and extraction of its
substantial oil, gas, and mineral deposits, scoring highly in the Natural Resource Governance
Institute’s Resource Governance Index. Brazil is also a founding member of the Open Government
Partnership, and has included several commitments related to extractives in its action plans.
Although a scoping study has been performed, Brazil is not yet a candidate for EITI membership.
Brazil is also one of the jurisdictions endorsing the OECD Declaration on Automatic Exchange of
Information in Tax Matters earlier this year. This unprecedented declaration will enable Brazil’s
tax authorities to collect information on Brazilian taxpayers’ overseas assets automatically,
and compare this data against their tax flings. Brazil is also a signatory to the Convention on
Mutual Administrative Assistance in Tax Matters and has bilateral relationships with many other
jurisdictions to exchange tax information upon request. While adoption of these instruments places
Brazil at the forefront of tax information sharing, the true test is how the information collected under
these arrangements is used, which is beyond our capacity to examine here.
iii. Financial Regulation and Governance
Brazil is a member of FATF and the Grupo de Acción Financiera de Sudamérica (GAFISUD), and
underwent a Mutual Evaluation in 2010,

which found Brazil’s laws to be largely or partially compliant
with most of the Recommendations.
32
However, signifcant holes remain—as noted, benefcial
ownership information is not available for all legal entities; legal entities cannot be subject to liability
for money laundering; and terrorist fnancing is not a distinct criminal offense, among others.
26.
Financial Action Task Force, Mutual Evaluation Report: Federative Republic of Brazil (Paris: FATF / OECD, 25 June 2010), 195, accessed
14 July 2014, http://www.fatf-gaf.org/countries/a-c/brazil/documents/mutualevaluationreportofbrazil.html.
27.
“2013 Resource Governance Index: Brazil,” Natural Resource Governance Institute, accessed 14 July 2014, http://www.
resourcegovernance.org/sites/default/fles/country_pdfs/brazilRGI2013.pdf.
28.
“Brazil,” Open Government Partnership, accessed July 14, 2014, http://www.opengovpartnership.org/country/brazil.
29.
See José Roberto Rodriguez Afonso, et al., Transparência Fiscal: Uma Análise da Indústria Extrativa Mineral Brasileira, World Bank, 10
October 2012. http://eiti.org/fles/Brazil%20Scoping%20Portuguese.pdf.
30.
“Countries commit to automatic exchange of information in tax matters,” Organization for Economic Co-Operation and Development, 6
May 2014, accessed 14 July 2014, http://www.oecd.org/tax/exchange-of-tax-information/countries-commit-to-automatic-exchange-of-
information-in-tax-matters.htm
31.
Global Forum on Transparency and Exchange of Information for Tax Purposes, “Brazil,” in Exchange of Tax Information Portal, accessed
14 July 2014, http://eoi-tax.org/jurisdictions/BR#agreements.
32.
FATF, Mutual Evaluation Report.
30 Global Financial Integrity
Furthermore, despite the strength of Brazil’s AML laws on paper, it is not clear that they are being
effectively enforced. The FATF Evaluation noted that despite Brazil’s high risks of money laundering
activity, the government pursued comparatively few investigations and obtained almost no criminal
convictions for money laundering. The Evaluation stated that this was likely due to structural and
logistical factors, such as short statutes of limitation, overtaxed courts, and limited prosecutorial
experience with complex fnancial cases, rather than a lack of motivation.
iv. Policy Recommendations
No set of policy changes is capable of completely eliminating illicit fnancial fows. Instead, the
goal should be to substantially curtail illicit fows,
33
through policies guided by two main principles:
greater transparency in domestic and international fnancial transactions, and greater cooperation
between developed and developing country governments to shut down the channels through which
illicit money fows. Brazil has already demonstrated a clear willingness to establish such principles,
through its commitments to open government and tax information exchange. In the paragraphs that
follow, we offer several key policy recommendations to guide the Government of Brazil in curtailing
future illicit fnancial fows.
a. Customs and Trade Reform
Addressing trade misinvoicing, the largest component of illicit fnancial fows out of Brazil, is a
complex undertaking, but the overall goal can be stated simply: ensure that goods being imported
or exported are recorded at a value based on the accurate market price of the goods. While this
necessarily requires greater vigilance on the part of customs inspectors and increased fexibility for
them to question transactions, they cannot bear full responsibility for it. And as noted above, Brazil
already gives its customs and tax inspectors signifcant ability to reconsider transaction values.
Instead, trade misinvoicing should be targeted from multiple angles, with an eye towards proactive
deterrence rather than retroactive punishment.
First, laws should be implemented specifcally criminalizing trade misinvoicing for the purposes of
evading taxes or tariffs, or to avoid money laundering controls. Additionally, importers and exporters
should be required to include and sign statements on declarations certifying that the prices stated
are accurate and honest. These simple steps could have a powerful deterrent effect through the
greater risk of detection and through the personal liability the declarations create.
Next, trade misinvoicing detection and identifcation should be incorporated into the generally
accepted accounting and auditing practices used in Brazil. Accountants and auditors of importing
or exporting businesses should be trained to identify red-fag transactions and verify whether they
were accurately invoiced. Both the executives and the auditors of Brazilian companies involved
33.
Raymond Baker et al., Hiding in Plain Sight: Trade Misinvoicing and the Impact of Revenue Loss in Ghana, Kenya, Mozambique,
Tanzania, and Uganda: 2002-2011 (Washington, DC: Global Financial Integrity, May 2014), 43.
31 Brazil: Capital Flight, Illicit Flows, and Macroeconomic Crises, 1960-2012
in international trade should be required to sign statements in the company’s annual accounts
certifying that all transactions included therein were invoiced in accordance with the law. As
with customs declarations, these simple statements would increase personal responsibility and
accountability for companies’ pricing decisions, deterring knowingly fraudulent conduct.
Finally, while Brazil has already taken steps towards considering misinvoicing as a risk category
for goods shipments and integrating pricing data into its processes, this should be expanded to
include goods in every Harmonized Code category. Customs inspectors should have access to
improved and expanded real-time world market pricing data, against which they can easily compare
the declared values of imports and request further documentation as needed. This data is now
becoming available from several sources.
b. Financial Transparency and Governance
Requiring legal entities registered in Brazil to disclose their benefcial owners—i.e., the natural
persons who ultimately control the company, regardless of the chain of ownership or legal authority
in between—is a powerful transparency measure that affects numerous problem areas related
to illicit fnancial fows. Laundering the proceeds of crime and corruption becomes much more
diffcult, hidden relationships between trading parties become much easier to identify, and banks’
customer due diligence requirements become substantially less onerous. Moves toward benefcial
ownership transparency in several major economies—the United Kingdom, France, and others—are
already underway. The central registry that Brazil already has in place should be augmented with a
legislative requirement for every registered company to list its benefcial owners, without regard to
the legal structure through which they control the company.
Brazil has already committed to joining the worldwide movement towards automatic exchange
of tax information, the “new global standard” as declared by the G20, and should now swiftly
look towards implementation of the system and effective utilization of the data to be gathered.
The OECD will be releasing a Commentary document later this year elaborating on the Common
Reporting Standard, and many countries will begin developing systems to collect and accept the
large amounts of data that will be required. The key to ensuring that this process meets Brazil’s
needs is collaboration: contributing to discussions on interpretation of the standard and developing
connections with other nations’ tax authorities. Building the technical and human capacity of RFB
will also be crucial to effectively utilizing the data.
Finally, Brazil should take the remaining steps necessary to fully implement the FATF
Recommendations and strengthen its anti-money laundering practices. While the government can
address much of the remaining concerns from the 2010 mutual evaluation with relatively minor
legislative amendments, addressing the structural factors needed to improve its implementation of
AML laws will require a much more holistic approach, including developing the capacity of the court
32 Global Financial Integrity
system and prosecutors and revising other procedural rules and regulations to handle complex
fnancial cases.
c. Effective Implementation
Overall, Brazil has an established fnancial infrastructure, a strong commitment to democratic
governance, and many of the laws and procedures needed to curb illicit fnancial fows and rein in
the underground economy already in place. However, these advantages must be coupled with the
capacity and political will to fully implement and enforce such measures. Curtailing illicit fnancial
fows must become a priority throughout the Brazilian government.

33 Brazil: Capital Flight, Illicit Flows, and Macroeconomic Crises, 1960-2012
V. Conclusion
The period 1960 to 2012 covered in this study saw massive structural changes in the Brazilian
economy: it evolved from one subject to various controls to a more market-based open economy.
Furthermore, over this 53-year period, Brazil experienced signifcant macroeconomic shocks such
as high and highly variable infation, hyperinfation, large fscal defcits, and crushing external debt
leading to debt default and deep recessions. This study analyzed the volume and pattern of both
broad capital fight and illicit fnancial fows from Brazil. While estimates of broad capital fight
were based on the World Bank Residual method adjusted for deliberate trade misinvoicing, illicit
fows were based on the Hot Money Narrow method, which was also similarly adjusted. We only
considered gross outfows and not a net of fows in both directions. As the Residual method as well
as the method of estimating illicit fows involve, either partly or wholly, capital that is illegally earned,
transferred, or utilized, netting out such fows would be methodologically unsound.
Over the 53-year period, Brazil lost a total of US$590.2 billion through broad capital fight, of which
US$401.6 billion was through illicit outfows. On average, these outfows represent 2.2 percent and
1.5 percent of GDP, respectively. The volume of capital fight increased exponentially from the 1960s
through the 1990s, although the pace declined over the last decade ending 2009. The continued
increase in capital fight in the 1990s has to do with outfows of licit capital in response to increasing
macroeconomic shocks such as hyperinfation and an onerous debt burden.
While both capital fight and illicit fows have tended to increase throughout the fve decades, they
tended to decline as a share of GDP. Starting at about 2.6 percent of GDP on average during the
1960s, capital fight fell slightly over the next two decades to touch 2.4 percent of GDP in the 1980s.
After that, the volume of capital fight increased back to 2.6 percent of GDP in the 1990s, before
outfows declined signifcantly to 1.9 percent of GDP in the last decade ending 2009. Over the last
three years, capital fight again increased slightly to 2.1 percent of GDP.
Outfows of illicit capital were around 1.5 percent of GDP in the 1960s and 1970s, increasing to 1.7
percent of GDP in the 1980s, before descending back to the vicinity of 1.4 to 1.5 percent of GDP in
the decades that followed.
We found that both capital fight and illicit outfows react predictably to macroeconomic shocks:
outfows seem to lead crises by a year or two, increase steadily throughout the period of economic
stress, and decline steadily in the aftermath. However, we found that the response of capital fight
to the “great recession” that started in early 2008 was more convincing than the behavior of illicit
outfows, which registered a plunge in the period 2010-2012.
An econometric model consisting of nine structural equations and one behavioral equation was
tested for the period 1965 to 2011. Six of the structural equations relate to the offcial economy and
34 Global Financial Integrity
three capture how broad capital fight, illicit fnancial fows, and the underground economy—which
we found to be 38.9 percent of the offcial economy on average per year of the study period—
interact with one another. Tests using the model showed that Brazil’s fscal policy did not play a
signifcant role in driving infation. Prices were mainly driven by increases in broad money supply.
While fscal defcits in the early 1960s and 1970s were fnanced through central bank credits and
money creation, bond fnancing together with foreign fnancing became much more important in the
2000s.
The model captured several aspects of the interaction between the above-ground, or offcial,
economy and the underground economy, illicit fows, and capital fight. On the one hand, nominal
income (GDP) was found to be a signifcant driver of investment (gross fxed capital formation). On
the other hand, growth of an underground economy, mainly driven by illicit fows, tended to divert
resources away from the offcial economy and had a signifcant negative impact on investment. In
other words, investment was being pushed by favorable developments in the offcial economy but
pulled back by growth of the underground economy, which was solely driven by illicit fows. Perhaps
the most signifcant fnding of the model developed in this study is that, while the underground
economy is mainly driven by illicit fows, broad capital fight was driven by governance-related
factors as well as macroeconomic drivers. Based on limited data, we found that worsening income
inequality also seems to drive capital fight, although the relationship is signifcant only at the 90
percent level.
The policy measures required to curtail capital fight and illicit fows are linked to the results of
the model simulations and guided by two main principles: 1) greater transparency in domestic
and international fnancial transactions and 2) greater cooperation between governments to
shut down the channels through which illicit money fows. These include taking stronger legal
measures against trade misinvoicing, instituting transparency of company ownership, and building
the technical and human capacity needed to effectively utilize the data that will be shared under
emerging tax information exchange arrangements.
Overall, Brazil has an established fnancial infrastructure, a strong commitment to democratic
governance, and many of the laws and procedures needed to curb illicit fnancial fows and rein in
the underground economy already in place. However, these advantages must be coupled with the
capacity and political will to fully implement and enforce such measures. Curtailing illicit fnancial
fows must become a priority throughout the Brazilian government.

35 Brazil: Capital Flight, Illicit Flows, and Macroeconomic Crises, 1960-2012
Appendix I. Capital Flight &
Illicit Financial Flows by Year
Table 1. Brazil: Broad Capital Flight and Illicit Financial Flows, 1960-2012
(in millions of U.S. dollars) 1/, 2/
Year
World Bank Residual
Method Outflows (WBR) (1)
Hot Money Outflows
(HMN) (2)
Trade Misinvoicing
Outflows (GER) (3)
Broad Capital Flight
(1+3)
Illicit Financial
Outflows (2+3)
1960
. 0 127 . 127
1961
. 0 153 . 153
1962
. 137 286 . 424
1963
. 77 302 . 378
1964
. 217 346 . 563
1965
699 31 230 929 261
1966
725 25 224 949 249
1967
1,236 35 263 1,498 297
1968
0 1 294 294 295
1969
0 41 309 309 350
1970
0 0 553 553 553
1971
225 7 659 884 666
1972
861 0 715 1,576 715
1973
0 0 1,015 1,015 1,015
1974
2,143 68 2,014 4,157 2,081
1975
604 438 3,015 3,619 3,453
1976
0 0 1,812 1,812 1,812
1977
4,816 628 2,106 6,922 2,734
1978
2,790 0 1,602 4,392 1,602
1979
1,370 0 3,599 4,969 3,599
1980
2,371 340 4,203 6,574 4,543
1981
0 418 4,002 4,002 4,421
1982
3,386 375 3,128 6,515 3,503
1983
0 586 3,013 3,013 3,600
1984
0 0 2,543 2,543 2,543
1985
1,463 530 4,111 5,574 4,641
1986
4,218 0 4,129 8,347 4,129
1987
9,392 805 4,379 13,770 5,184
1988
3,099 827 6,016 9,115 6,842
1989
0 819 6,487 6,487 7,307
1990
1,380 296 6,799 8,179 7,096
1991
35 0 6,414 6,449 6,414
1992
1,543 1,393 5,987 7,530 7,380
1993
7,175 815 5,916 13,091 6,730
1994
1,970 442 7,022 8,992 7,464
1995
0 0 7,681 7,681 7,681
1996
918 1,992 7,336 8,254 9,328
1997
13,588 3,160 10,178 23,766 13,338
1998
45,379 2,911 9,072 54,451 11,983
1999
12,358 0 8,306 20,663 8,306
2000
5,930 0 9,430 15,360 9,430
2001
0 498 9,761 9,761 10,259
2002
8,142 154 8,832 16,974 8,986
2003
9,556 933 11,221 20,777 12,153
2004
2,981 2,145 13,856 16,837 16,001
2005
0 225 16,716 16,716 16,941
2006
0 0 10,805 10,805 10,805
2007
0 3,152 14,347 14,347 17,499
2008
17,947 0 22,375 40,321 22,375
2009
0 0 22,237 22,237 22,237
2010
11,260 3,559 29,001 40,261 32,560
2011
8,255 1,274 33,042 41,296 34,316
2012
31,380 0 34,286 65,666 34,286
Cumulative
219,194 29,353 372,254 590,234 401,607
Average
4,567 554 7,024 12,297 7,577
1/ All U.S. dollar fgures are nominal
2/ (.) indicates no available data, whereas (0) indicates a value of 0.
36 Global Financial Integrity
Appendix II. The Components of Brazil’s Trade Misinvoicing
Table 2. Brazil: The Components of Trade Misinvoicing, 1960-2012
(in millions of U.S. dollars) 1/, 2/
Year
Import Misinvoicing Export Misinvoicing
Total Inflows
(b+c)
Total Outflows
(a+d)
Gross Trade
Misinvoicing
(a+b+c+d)
Over-Invoicing
(a)
Under-Invoicing
(b)
Over-Invoicing
(c)
Under-Invoicing
(d)
1960
43 78 41 84 119 127 246
1961
52 90 91 101 181 153 334
1962
96 25 50 191 75 286 362
1963
146 12 107 155 119 302 421
1964
124 21 52 221 74 346 419
1965
65 54 62 165 116 230 346
1966
32 41 53 192 95 224 319
1967
72 60 64 191 124 263 387
1968
38 139 48 256 187 294 481
1969
104 59 126 205 185 309 495
1970
263 26 142 290 167 553 721
1971
316 6 151 343 157 659 816
1972
352 53 246 363 299 715 1,014
1973
387 128 447 628 576 1,015 1,591
1974
1,359 219 576 654 795 2,014 2,809
1975
1,882 0 387 1,133 387 3,015 3,402
1976
1,382 6 949 430 955 1,812 2,767
1977
972 401 1,017 1,135 1,418 2,106 3,524
1978
930 127 825 672 952 1,602 2,554
1979
1,718 275 654 1,881 929 3,599 4,528
1980
2,241 103 979 1,962 1,082 4,203 5,285
1981
1,727 90 1,831 2,275 1,921 4,002 5,923
1982
997 940 1,000 2,131 1,940 3,128 5,069
1983
1,300 875 1,465 1,714 2,341 3,013 5,354
1984
297 1,727 1,418 2,246 3,145 2,543 5,688
1985
463 1,159 1,228 3,648 2,387 4,111 6,498
1986
492 1,340 1,152 3,637 2,491 4,129 6,620
1987
611 1,969 1,388 3,768 3,357 4,379 7,736
1988
541 2,278 2,327 5,475 4,606 6,016 10,621
1989
1,027 2,065 2,428 5,461 4,493 6,487 10,980
1990
1,543 1,639 2,076 5,256 3,715 6,799 10,515
1991
1,287 1,886 1,810 5,127 3,696 6,414 10,109
1992
300 1,660 2,406 5,686 4,066 5,987 10,053
1993
1,946 914 3,677 3,970 4,591 5,916 10,506
1994
1,829 3,395 3,506 5,193 6,901 7,022 13,923
1995
1,552 4,904 3,137 6,129 8,041 7,681 15,722
1996
1,695 4,477 5,325 5,642 9,801 7,336 17,138
1997
2,600 6,494 5,070 7,578 11,563 10,178 21,741
1998
2,222 6,298 2,632 6,850 8,930 9,072 18,002
1999
2,712 4,446 1,702 5,594 6,148 8,306 14,454
2000
2,410 6,655 2,206 7,020 8,861 9,430 18,291
2001
2,522 8,188 1,884 7,238 10,072 9,761 19,833
2002
2,063 7,423 1,776 6,769 9,199 8,832 18,031
2003
2,136 5,906 1,748 9,085 7,653 11,221 18,874
2004
1,999 9,529 2,744 11,857 12,274 13,856 26,130
2005
2,643 10,870 2,239 14,073 13,108 16,716 29,824
2006
2,716 13,896 4,353 8,089 18,249 10,805 29,054
2007
4,674 18,948 7,059 9,673 26,008 14,347 40,355
2008
5,603 23,853 7,108 16,772 30,960 22,375 53,335
2009
4,662 19,066 9,847 17,574 28,913 22,237 51,150
2010
10,996 23,147 14,675 18,005 37,821 29,001 66,822
2011
13,533 28,971 18,624 19,509 47,595 33,042 80,637
2012
11,520 32,258 24,688 22,766 56,946 34,286 91,232
Cumulative
105,191 259,188 151,596 267,063 410,784 372,254 783,038
Average
1,985 4,890 2,860 5,039 7,751 7,024 14,774
1/ All U.S. dollar fgures are nominal
2/ (.) indicates no available data, whereas (0) indicates a value of 0.
37 Brazil: Capital Flight, Illicit Flows, and Macroeconomic Crises, 1960-2012
Appendix III. Illicit Flows to GDP and Trade
Table 3. Brazil: Illicit Flows to GDP and Trade, 1960-2012
(in millions of U.S. dollars or as percent) 1/, 2/
Year Illicit Financial Flows GDP Total Trade
Illicit Financial Flows
to GDP
Illicit Financial Flows
to Total Trade
1960
127 15,166 2,729 0.84% 4.65%
1961
153 15,237 2,863 1.00% 5.34%
1962
424 19,926 2,688 2.13% 15.78%
1963
378 23,021 2,862 1.64% 13.21%
1964
563 21,212 2,665 2.65% 21.13%
1965
261 21,790 2,667 1.20% 9.79%
1966
249 27,063 3,206 0.92% 7.77%
1967
297 30,592 3,284 0.97% 9.04%
1968
295 33,876 3,968 0.87% 7.43%
1969
350 37,459 4,531 0.93% 7.72%
1970
553 42,328 5,532 1.31% 10.00%
1971
666 49,204 6,608 1.35% 10.08%
1972
715 58,539 8,774 1.22% 8.15%
1973
1,015 79,279 13,202 1.28% 7.69%
1974
2,081 105,000 22,117 1.98% 9.41%
1975
3,453 124,000 22,309 2.78% 15.48%
1976
1,812 153,000 23,943 1.18% 7.57%
1977
2,734 176,000 25,383 1.55% 10.77%
1978
1,602 201,000 27,743 0.80% 5.77%
1979
3,599 225,000 34,901 1.60% 10.31%
1980
4,543 235,000 45,124 1.93% 10.07%
1981
4,421 264,000 47,370 1.67% 9.33%
1982
3,503 282,000 41,242 1.24% 8.49%
1983
3,600 203,000 38,750 1.77% 9.29%
1984
2,543 209,000 42,215 1.22% 6.02%
1985
4,641 223,000 39,972 2.08% 11.61%
1986
4,129 268,000 37,990 1.54% 10.87%
1987
5,184 294,000 42,539 1.76% 12.19%
1988
6,842 330,000 49,575 2.07% 13.80%
1989
7,307 426,000 54,267 1.72% 13.46%
1990
7,096 462,000 53,936 1.54% 13.16%
1991
6,414 407,000 54,567 1.58% 11.75%
1992
7,380 391,000 58,909 1.89% 12.53%
1993
6,730 438,000 66,159 1.54% 10.17%
1994
7,464 614,000 79,737 1.22% 9.36%
1995
7,681 769,000 100,644 1.00% 7.63%
1996
9,328 840,000 104,728 1.11% 8.91%
1997
13,338 871,000 117,237 1.53% 11.38%
1998
11,983 844,000 111,792 1.42% 10.72%
1999
8,306 587,000 99,779 1.41% 8.32%
2000
9,430 645,000 113,762 1.46% 8.29%
2001
10,259 554,000 116,668 1.85% 8.79%
2002
8,986 504,000 110,161 1.78% 8.16%
2003
12,153 552,000 124,084 2.20% 9.79%
2004
16,001 664,000 163,111 2.41% 9.81%
2005
16,941 882,000 196,157 1.92% 8.64%
2006
10,805 1,090,000 233,645 0.99% 4.62%
2007
17,499 1,370,000 287,294 1.28% 6.09%
2008
22,375 1,650,000 380,320 1.36% 5.88%
2009
22,237 1,620,000 286,667 1.37% 7.76%
2010
32,560 2,140,000 393,452 1.52% 8.28%
2011
34,316 2,480,000 492,985 1.38% 6.96%
2012
34,286 2,253,090 470,957 1.52% 7.28%
1/ All U.S. dollar fgures are nominal
2/ (.) indicates no available data, whereas (0) indicates a value of 0.
38 Global Financial Integrity
Appendix IV. Estimating Brazil’s
Underground Economy
Measuring the informal, or underground, sector of an economy has been of interest to many
researchers concerned with development. There are primarily three categories of techniques used
to measure for informality.
1. Direct methods: methods which involve taking public surveys and conducting interviews with
actual informal workers.
2. Indirect methods: methods in which discrepancies in offcial records are used as proxies to
obtain the size of the informal sector.
3. Multiple Indicators Multiple Causes (MIMIC) approach: as made popular by Schneider, MIMIC
models aim to link unobserved variables to observed ones to derive the size of the underground
economy.
34

Due to data constraints, we model our estimates of the underground economy in Brazil according
to the Currency Demand approach, which falls under the “indirect method” category of techniques.
This has been the approach of many studies on informality, and was pioneered by the works of
Tanzi.
35
We model our estimates very similar to Tanzi’s, but along the lines of Macias due to data
limitations and the issues related to using the ratio of currency demand to holdings of money.
36
Our
fnal model was as follows:
Ct = β
0
+ β
1
+ β
2
Tax
t
- β
t
IR
t
Where C is the currency held outside banks normalized by the price level, Y is real income, Tax
represents total tax revenue collected, and IR is the nominal effective interest rates.
Due to the presence of non-stationarity and cointegration in all the variables involved, we use
a vector error correction model (VECM) to model the above equation. The coeffcients are then
normalized around C to obtain the long-run equation. The crux of the currency demand approach
lies in comparing what currency holdings outside depository institutions would be if the tax rate
were to fall to zero, assuming that taxes are one of the chief causes of individuals remaining in the
informal sector. The difference between the above model estimated with taxes and without taxes
34.
Friedrich Schneider, “Measuring the Size and Development of the Shadow Economy. Can the Causes be Found and the Obstacles be
Overcome?,” in Essays on Economic Psychology, eds. Hermann Brandstaetter and Werner Güth (Berlin: Springer Publishing Company,
1994a).
35.
Vito Tanzi, The Underground Economy in the United States: Annual Estimates, 1930-80, IMF Staff Papers 30(2) (Washington, DC: IMF,
June 1983).
36.
Jose Brambila Macias, The Dynamics of Parallel Economies. Measuring the Informal Sector in México, MPRA Paper No. 8400 (Munich:
University Library of Munich, 2008), 4.
39 Brazil: Capital Flight, Illicit Flows, and Macroeconomic Crises, 1960-2012
gives us an estimate of the extra currency in the economy. This fgure is then multiplied by the
velocity of money, similar to Tanzi’s and numerous other studies, to get our fnal estimate.
37

Our estimates put the average size of the underground economy to GDP at 38.9 percent over the
entire period of study. This puts our estimates similar to, but slightly less, than Schneider et al.’s
estimate of 39.0 percent for the period 1999-2007.
38

Chart 4. Underground Economy in Brazil, Decadal Averages, 1960-2012
(in percent of GDP)
0%
10%
20%
30%
40%
50%
60%
1960-1969 1970-1979 1980-1989 1990-1999 2000-2009 2010-2012
P
e
r
c
e
n
t

o
f

G
D
P

Underground Economy to GDP
Table 4. Underground Economy in Brazil, Decadal Averages, 1960-2012
(in percent of GDP)
Year
Average Underground
Economy to GDP
1960-1969
45.76%
1970-1979
55.09%
1980-1989
51.78%
1990-1999
36.30%
2000-2009
33.27%
2010-2012
21.79%
1960-2012
38.90%
37.
Tanzi, Underground Economy.
38.
Friedrich Schneider, Andreas Buehn, and Claudio E. Montenegro, “New Estimates for the Shadow Economies all over the World,”
International Economic Journal 24:4 (2010), 454.
40 Global Financial Integrity
41 Brazil: Capital Flight, Illicit Flows, and Macroeconomic Crises, 1960-2012
References
Afonso, José Roberto Rodrigues, Julia Morais Soares, Kleber Pacheco de Castro, and Ricardo
Figueiró Silveira. Transparência Fiscal: Uma Análise da Indústria Extrativa Mineral Brasileira.
World Bank, 10 October 2012. Accessed 14 July 2014, http://eiti.org/fles/Brazil%20
Scoping%20Portuguese.pdf.
Baker, Raymond, Christine Clough, Dev Kar, Brian LeBlanc, and Joshua Simmons. Hiding in Plain
Sight: Trade Misinvoicing and the Impact of Revenue Loss in Ghana, Kenya, Mozambique,
Tanzania, and Uganda: 2002-2011. Washington, DC: Global Financial Integrity, May 2014.
Barnato, Katy. “This Nation Could Be the Most at Risk From Capital Flight.” CNBC, 14 June 2013.
accessed 13 May 2014, http://www.cnbc.com/id/100815904.
Claessens, Stijn, and David Naudé. Recent Estimates of Capital Flight. Policy Research Working
Paper Series 1186. Washington, DC: Debt and International Finance Division, International
Economics Department, World Bank, 1993.
Cuddington, John T. “Capital Flight: Estimates, Issues, and Explanations.” In Princeton Studies
in International Finance. Princeton, NJ: International Finance Section, Dept. of Economics,
Princeton University, 1986.
Cumby, Robert, and Richard Levich. On the Defnition and Magnitude of Recent Capital Flight. In
Capital Flight and Third World Debt, edited by D. Lessard and J. Williamson. Washington, DC:
Institute for International Economics, 1987.
Deppler, Michael, and Martin Williamson. “Capital Flight: Concepts, Measurement, and Issues.”
In Staff Studies for the World Economic Outlook, SM/87/24. Washington, DC: International
Monetary Fund, 23 January 1987.
Dooley, Michael P. Capital Flight: A Response to Differences in Financial Risks. IMF Staff Papers 35,
422-436. Washington, DC: International Monetary Fund, 1988.
Financial Action Task Force. Mutual Evaluation Report: Federative Republic of Brazil. Paris: FATF /
OECD, 14 July 2014. Accessed 14 July 2014, http://www.fatf-gaf.org/media/fatf/documents/
reports/mer/MER%20Brazil%20full.pdf.
Galvao, Arnaldo. “Brazil May Face Capital Flight on European Debt, IMF Director Says.” Bloomberg,
17 October 2011, accessed 16 May 2014, http://www.bloomberg.com/news/2011-10-17/brazil-
may-face-capital-fight-on-european-debt-crisis-imf-director-says.html.
42 Global Financial Integrity
Global Forum on Transparency and Exchange of Information for Tax Purposes. “Brazil.” In
Exchange of Tax Information Portal. Accessed 14 July 2014, http://eoi-tax.org/jurisdictions/
BR#agreements.
International Monetary Fund. “Fiscal Sustainability and Monetary Versus Fiscal Dominance:
Evidence from Brazil, 1991-2000.” In Brazil: Selected Issues and Statistical Appendix, IMF
Country Report No. 01/10. Washington, DC: IMF, January 2001.
—. “Brazil: Staff Report for the 2012 Article IV Consultation.” In Brazil: 2012 Article IV Consultation –
Staff Report; Public Information Notice on the Executive Board Discussion; and Statement by
the Executive Director for Brazil, IMF Country Report No. 12/191. Washington, DC: IMF, July
2012.
Kar, Dev. “The Brazilian Financial Sector: An Empirical Study in Evolution.” PhD diss., George
Washington University, 1982.
—. Government Deficits and Inflation in Brazil: The Experience During 1948-64. IMF Working Paper
DM/81/76. Washington, DC: International Monetary Fund, October 1981.
Kar, Dev, and Brian LeBlanc. Illicit Financial Flows from Developing Countries: 2002-2011.
Washington, DC: Global Financial Integrity, 2013.
Natural Resource Governance Institute. “2013 Resource Governance Index: Brazil.” Accessed 14
July 2014, http://www.resourcegovernance.org/sites/default/fles/country_pdfs/brazilRGI2013.
pdf.
Open Government Partnership. “Brazil.” Accessed 14 July 2014, http://www.opengovpartnership.
org/country/brazil.
Organization for Economic Cooperation and Development. “Countries commit to automatic
exchange of information in tax matters.” 5 June 2014. Accessed 14 July 2014, http://www.
oecd.org/tax/exchange-of-tax-information/countries-commit-to-automatic-exchange-of-
information-in-tax-matters.htm.
PwC. “Brazil.” In International Transfer Pricing 2013/14, 284-97. New York: PwC, 2013. Accessed 14
July 2014, http://www.pwc.com/gx/en/international-transfer-pricing/assets/itp-2013-fnal.pdf.
—. “Brazil.” In Worldwide Tax Summaries: Corporate Taxes 2013/14, 264-280. New York: PwC,
2013. Accessed 14 July 2014, http://www.pwc.com/gx/en/tax/corporate-tax/worldwide-tax-
summaries/assets/pwc-worldwide-tax-summaries-corporate-2013-14.pdf.
43 Brazil: Capital Flight, Illicit Flows, and Macroeconomic Crises, 1960-2012
—. “Brazilian Federal Revenue Department issues further guidance on changed introduced by
Law 12715,” Tax Insights, January 9, 2013, http://www.pwc.com/en_GX/gx/tax/ newsletters/
pricing-knowledge-network/assets/pwc-brazil-law12715-guidance.pdf.
Schneider, Friedrich. “Measuring the Size and Development of the Shadow Economy: Can the
Causes be Found and the Obstacles be Overcome.” In Essays on Economic Psychology,
edited by Hermann Brandstaetter and Werner Güth, 193-212. Berlin: Springer Publishing
Company, 1994.
Schneider, Freidrich, Andreas Buehn, and Claudio E. Montenegro. “New Estimates for the Shadow
Economies all over the World.” International Economic Journal 24:4 (2010), 443-461.
Stuber, Walter, and Adriana Maria Gödel Stuber, “Brazil: Brazilian Tax Authorities Issue a New List
of Favored Taxation Countries.” Mondaq, June 10, 2010. Accessed 3 July 2014, http://www.
mondaq.com/ x/102618/Income+Tax/Brazilian+Tax+Authorities+Issue+a+New+List+of+Favore
d+Taxation+Countries.
Tanzi, Vito. The Underground Economy in the United States: Annual Estimates, 1930-1980. IMF Staff
Papers 30(2). Washington, DC: International Monetary Fund, June 1983.
Transparency International. “Corruption by Country: Brazil.” Accessed 14 July 2014, http://www.
transparency.org/country#BRA.
World Trade Organization. Trade Policy Review: Brazil, WT/TPR/S/283. Geneva: WTO, 17 May 2013.
Accessed 14 July 2014, http://www.wto.org/english/tratop_e/tpr_e/tp383_e.htm.
44 Global Financial Integrity
About Global Financial Integrity
Global Financial Integrity (GFI) is a non-proft, Washington, DC-based research and advocacy
organization, which produces high-caliber analyses of illicit fnancial fows, advises developing
country governments on effective policy solutions, and promotes pragmatic transparency measures
in the international fnancial system as a means to global development and security.
About the Author
Dev Kar is the Chief Economist at Global Financial Integrity. Prior to joining GFI, Dr. Kar was
a Senior Economist at the International Monetary Fund (IMF), Washington DC. During a career
spanning nearly 32 years at the IMF, he worked on a wide variety of macroeconomic and statistical
issues, both at IMF headquarters and on different types of IMF missions to member countries
(technical assistance, Article IV Consultations with member countries, and Use of IMF Resources).
He has published a number of articles on macroeconomic and statistical issues both inside and
outside the IMF. Dr. Kar has a Ph.D. in Economics (Major: Monetary Economics), an M. Phil in
Economics (Major: International Economics) from the George Washington University, and an
M.S. in Computer Science (Major: Database Management Systems) from Howard University. His
undergraduate degree in Physics is from St. Xavier’s College, University of Calcutta, India.
1100 17th Street, NW, Suite 505 | Washington, DC | 20036 | USA
Tel. +1 (202) 293-0740 | Fax. +1 (202) 293-1720 | www.gfntegrity.org
President: Raymond Baker Managing Director: Tom Cardamone
Board: Lord Daniel Brennan (Chair), Dr. Rafael Espada (Vice Chair),
Dr. Lester A. Myers (Secretary-Treasurer), Dr. Thomas Pogge, Raymond Baker

Sign up to vote on this title
UsefulNot useful

Master Your Semester with Scribd & The New York Times

Special offer for students: Only $4.99/month.

Master Your Semester with a Special Offer from Scribd & The New York Times

Cancel anytime.